With every major broadcaster launching over-the-top (OTT) offerings and independent players multiplying, the CTV advertising market is getting critical traction.

As of mid-2025, streaming accounted for 44.8% of total TV viewership, surpassing the combined share of broadcast (20.1%) and cable (24.1%) for the first time in history.

CTV ad spending is set to grow from $33.35 billion in 2025 to $46.89 billion by 2028, when it will surpass traditional TV ad spending ($45.10 billion) for the first time, according to eMarketer

However, media buyers are right to have mixed feelings about CTV advertising. 

The lack of transparency and proper safeguards in CTV costs advertisers an average of $700,000 in wasted spend per billion impressions. 

Advertisers point out that it’s difficult to tell whether CTV buys are reaching viewers due to the highly fragmented ecosystem. A DoubleVerify report found that only 50% of all CTV impressions offer full transparency, and even so, CTV advertising is still perceived as difficult to measure. 

Fortunately, connected TV ads can provide data points as relevant as those from other digital channels with a proactive approach to partnerships and interoperability. 

In this post, you’ll learn about:

• The fragmented CTV market landscape and its implications for AdTech companies 
• The main challenges of CTV advertising measurement and attribution 
• Best tech practices for gaining CTV measurement data that buyers need 

CTV market overview: Platforms & operating systems (OS)  

The CTV market is an ecosystem. Participants include smart TV device manufacturers, standalone media players, OTT providers, and content distribution platforms. All of them have a heavy hand in the market because they own (but do not always share) consumer data. 

To gain full visibility into CTV ad performance, ad platforms have to integrate data from multiple sources. What makes CTV measurement even harder is that no single player dominates the smart TV OS market or the OTT market.  

Main types of CTV players 

• Smart TVs with native OS (e.g., Samsung TV, LG TV, Sony, Vizio with embedded Chromecast) 
• Stand-alone streaming devices and media players ( e.g., Roku, Amazon Fire, Chromecast, or Apple TV) 
• OTT video-streaming services (e.g., AT&T TV, HBO Max, Hulu, Netflix, Paramount+, Rakuten TV, etc.)
• Content distribution platforms (e.g., Amagi, Castify.ai, BitCentral, Viaccess-Orca, etc.) 

That said, the global CTV market has its “big four” players, holding most of the audience data (and advertising dollars). 

Samsung Connected TV 

Samsung was among the first to release competitively priced smart TV sets. Since its market launch in 2015, the installed base of Samsung Tizen has grown to 270 million TV and smart signage devices worldwide. 

On a global scale, Samsung remains a leader, though the competitive landscape has shifted significantly. Android/Google TV is now the leading Smart TV OS, accounting for over 24% of global shipments, with Tizen at 16.9%, WebOS at 11.8%, and Roku at 9%. 

Hisense’s VIDAA OS has emerged as a major competitor at 7.8% global market share, followed by LG WebOS at 7.4%, with Roku and Amazon Fire TV tied at 6.4%. However, Samsung continues to trail in the North American market, where Roku leads the CTV device market share at 37%, followed by Amazon Fire TV at 17%, while Samsung holds just 12%.

Roku 

The first Roku streaming device was released with Netflix in 2008. Since then, the company has expanded its hardware product range, developed the Roku OS, and launched a programmatic CTV advertising network.

Roku reached more than 90 million streaming households as of the first week of January 2025, making it an attractive platform for OLV advertising. Roku’s Platform revenue surpassed $1 billion for the first time in Q4 2024, growing 25% year-over-year. In the Q4 2024 earnings call, Roku’s CEO noted that at least one Roku-powered device is in half of US broadband homes.

However, Roku’s devices segment faced challenges with a full-year 2024 gross margin of -14% and a Q4 gross margin of -29% due to increased seasonal discounts.

Amazon Fire TV 

Amazon entered the CTV space with affordable Fire sticks, went on to launch Fire TV (an edition of smart television sets), signed Fire OS distribution deals with popular device manufacturers (Insignia, Toshiba, JVC, Grundig, and, more recently, Panasonic). 

To date,  Amazon has sold more than 250 million Fire TV devices globally since the platform’s launch in 2014, with an increase of 50 million since late 2023

Amazon has also been exploring the emerging in-car video streaming market. At CES 2022, Amazon announced a pact with Ford Motor Co. to embed Fire TV in Ford Expedition and Lincoln Navigator models, and separately announced a deal with Stellantis to integrate Fire TV into Wagoneer, Grand Wagoneer, Jeep Grand Cherokee, and Chrysler Pacifica models.

Google TV (Android TV)

Google entered the connected TV space with Chromecast devices (smart TV sticks), but quickly assembled a larger ecosystem of products. The Android TV platform is the original Google OS for smart TV sets.

In 2020, Google released a major upgrade to Android TV and rebranded its offering as Google TV. At its core, Google TV is a new interface running on top of the original Android TV OS. 

It comes pre-installed on the Google TV Streamer (which replaced the Chromecast line in 2024) and is the primary interface for smart TV manufacturers that opted for Android TV OS. 

Google is progressively phasing out the older Android TV interface in favor of Google TV across all devices. Google TV now comes pre-installed on smart TVs from brands like TCL, Sony, Hisense, Sharp, Philips, and others. As of September 2024, Google TV is active on over 270 million devices monthly

What CTV market fragmentation means for the AdTech Industry

Device and data fragmentation is the bane of all new channels, like in-game advertising or DOOH. Sourcing data from multiple smart TV sets, OTT providers, and OS is technically complex. In addition to many conflicting requirements and limitations is a lack of standardization. Combined, these factors complicate CTV ad measurement.

On the other hand, as Tal Chalozin, CTO and Co-Founder at Innovid, an independent CTV measurement platform, rightfully noted: 

Fragmentation means competition, and competition means lower prices. When platforms have to compete against one another to secure ad dollars, then the number one lever available to them is their price. As long as the connected TV space remains heavily fragmented, marketers will benefit from a buyer’s market.

More advertisers consider CTV advertising. AdTech companies that can develop better CTV ad measurement solutions and provide precise attribution metrics will emerge on top. 

CTV advertising measurement challenges

CTV attribution is hard primarily due to the absence of shared standards for measurability.

Back in the day, Nielsen pioneered measurement for linear TV advertising. Though the company made a tentative move into CTV measurement, both of its frameworks are often criticized for inaccurate audience counts. 

Brands (and their agency partners) are on the hunt for a better measurement solution. Which one will it be? The following could resolve the CTV measurement and attribution issues. 

Lack of common identifiers

The digital advertising space relied on third-party cookies for years to identify, track, and report user behaviors. Now the industry works towards universally acceptable cookieless tracking and shared user ID solutions.

CTV ad space faces a similar dilemma: It needs cross-platform identifiers. IP addresses have been the most common means of identifying households as they are easy to capture. Most programmatic CTV advertising uses IP addresses for targeting and remarketing. 

But is an IP address a reliable ID? No. Many consumers share streaming accounts and use various devices to view the content (i.e., the IP address changes, but the user stays the same or vice versa). Because neither supply-side platforms (SSPs) nor demand-side platforms (DSPs) can precisely ID users, a lot of budgets are wasted. For example, if a brand buys connected TV ads through Roku and via a DSP platform, they risk marketing ad duplication. According to the Innovid x ANA Report: 

The average CTV campaign frequency was 7.09 in 2024, with an average CTV household reach of only 19.64%. As campaign sizes grow, so does the risk of oversaturation: high-investment campaigns with over 200M+ impressions saw frequency rise to 10+.

So what are the good options? CTV-specific user identity graphs may help. Digital ID providers like Ramp ID (former IdentityLink) and Tapad offer connected TV capabilities as part of omnichannel identity graphs. However, both solutions primarily rely on IP addresses for initial user identification. Then they augment the created identity with other data points.

No viable alternatives to IP addresses have been found so far, apart from first-party-based ID solutions built by different players in the ecosystem. That said, IP addresses aren’t definitely going away just yet. So the industry has time to come up with new ID types like device graphs or universal user ID graphs. 

Multitude of different CTV measurement methodologies

When you ask Ad Ops which CTV measurement metrics they use, you’ll get an entire spreadsheet of answers: 

Buyers want both familiar linear TV metrics and programmatic ones. Yet, many DSPs and SSPs struggle to deliver such a large roster of accurate insights. So brands are eager to test multiple CTV attribution options on the table. The Trade Desk and Viant Technology already went with iSpot. Xandr, ABEMA, Smadex, and tvScientific have selected Adjust. 

Why do brands want multiple partners? Because the “big four” CTV platforms (Samsung, Roku, Amazon, and Google) employ proprietary approaches to measurement (which they don’t fully disclose).

While Nielsen has expanded into CTV measurement, its cross-platform coverage is still evolving, leaving gaps in independent verification.

Also, fragmentation exists on the AdTech level, where buyers can purchase CTV ads via different ad platforms directly. This further splinters audience data and complicates measurement.

Complex device identification process 

Since most platforms rely on IP addresses for user identification, it’s hard to determine who saw the ad: the same person on two different devices, multiple people on one device, or multiple people via the same OTT app. 

Also, CTV/OTT ads rely on the server-side ad insertion (SSAI) mechanism. It seamlessly integrates ad videos into the streamed content. SSAI is resistant to ad blockers and allows low-latency ad serving. However, SSAI needs accurate device ID data to deliver accurate impression counts. 

IAB Tech Lab’s original 2019 guidelines for CTV/OTT device and app identification recommended using “app store IDs” where available, but significant challenges persist. A lack of standardization around the syntax of Bundle IDs has led to confusion around targeting and measurement, creating a vulnerability that fraudsters could exploit.

To address these persistent identification challenges, IAB Tech Lab created the Ad Creative ID Framework (ACIF) in 2024 to simplify ad creative management and tracking across platforms. It supports the use of registered creative IDs that persist in cross-platform digital video delivery, particularly in CTV environments. The ACIF Validation API entered public comment in December 2024, and ACIF Version 1.0 was released in March 2025.

Using the WURFL device detection database is one workaround. It streamlines user device identification (device model, browser, OS, screen width, etc.). WURFL can be used to improve CTV attribution when paired with machine learning. Still, the setup process is quite complex. 

Cross-media measurement

Market fragmentation means that consumers have a lot of choices. Naturally, most switch between watching linear TV, using CTV apps, and OTT services on mobile. 

The wrinkle? Few exchange data with one another. Audience data is siloed between:

• Digital multichannel video programming distributors (MVPDs) 
• Direct-to-consumer OTT apps
• Smart TV manufacturers
• CTV OS distributors 
• SSPs, DSPs, and ad networks 

As a result, procuring data points such as device ID, audience demographic, or average viewership is hard, even for original content owners. Distributors typically hold most of the data to attract demand, though some publishers now buy back audience insights. Getting a consolidated view of video content viewership rates is somewhat problematic. 

CTV advertising fraud 

Programmatic ad fraud is a gruesome industry issue. CTV ads are no exception. 

Complex attribution stands behind high IVT rates in CTV advertising. Because verified data is hard to produce, faking ad impressions for CTV is easier than for desktop or mobile devices (although sophisticated ad fraud detection mechanisms might help).

Organizations like IAB Open Measurement, Media Rating Council (MRC), Trustworthy Accountability Group (TAG), and Brand Safety Institute have released comprehensive CTV ad fraud prevention guidelines. The challenge, however, lies in implementing them. 

6 best practices of CTV measurement 

No single metric can indicate the success of a CTV ad campaign. To reassure the buy-side, AdTech players have to provide a roster of cross-channel metrics, proving ad validity and viewability. 

Of course, the best industry minds are working on the CTV measurement problem. In May 2024, IAB Tech Lab expanded its Open Measurement SDK (OM SDK) to include Samsung and LG platforms, now covering 40% of CTV households.

The framework continues to evolve as a common standard for interoperability, with IAB Tech Lab releasing Device Attestation support in late 2025 to combat device spoofing in CTV environments.

OM SDK gives advertisers flexibility and choice in the verification solutions from their preferred providers by making it easier for publishers to integrate one SDK and enable ad verification with all verification vendors.

The IAB Tech Lab announcement

OM SDK is a helpful tool, but not a stand-alone solution. To improve CTV measurement, you need to combine several best practices. 

Employ a hybrid approach to cross-channel attribution 

Because access to audience data is constrained, no best-of-breed user attribution solution is available. Instead, the industry tests various methods for identifying users and tracking their interactions with content.

IAB suggests that the path pass forward would be using hybrid measurement approaches that combine:

• Automatic content recognition (ACR) methods, such as audio fingerprinting or watermarking 
• Passive panel metering technologie,s such as people meters 
• Digital metering using linked mobile devices or home router-level meters
• Third- or first-party census feeds

The combination of these signals can enable industry players to minimize ad duplication and better distinguish between linear TV, CTV app feeds at the household and individual levels, and broadcast video on demand (BVOD). 

Separately, user ID data such as identifiers for advertising (IFAs), CTV IDs, device IDs, and IP addresses could be cross-matched with audience profiles across platforms. In fact, most market players are making strides in this direction. 

Verizon Media ID 

Yahoo DSP (formerly Verizon Media) ConnectID includes CTV household data. In 2021, the company partnered with smart TV manufacturer VIZIO to gain viewership data from some 18 million VIZIO Smart TVs. 

However, the CTV landscape has shifted significantly since then, and Walmart acquired VIZIO in 2024. Now, one of the largest US retailers’ ecosystems is linked with a major source of TV viewership data, creating new opportunities for retail media targeting on CTV.

Roku Advertising Watermark

In early 2022, Roku released Advertising Watermark, a platform-native way to validate video ads’ authenticity on the Roku platform. The technology has since evolved significantly: in 2023, Roku launched Watermark 2.0, which detects fake impressions at both the device and app level and can be passed through the programmatic bidstream. 

Working with partners like DoubleVerify and HUMAN, the watermark has helped combat major fraud schemes, including CycloneBot, which generated up to 250 million fake ad requests daily.

Roku reports a marked reduction in fraudulent ad requests imitating its device traffic since 2023. The watermark is now integrated with Roku Ads Manager, which has replaced OneView as Roku’s primary ad-buying platform.

Determine the optimal approach to audience measurement

Since CTV is a cookieless environment, precise audience measurement is complex but possible. The Media Rating Council (MRC) has an exhaustive list of standards and approaches to cross-media CTV audience measurement. 

In short, there are two main options:

• pixel-based technology to capture an impression, video start, and completion data; and to detect and report on Invalid Traffic.
• embedded SDK or client-side measurement code for cross-channel measurement (such as OM SDK by IAB).

Once again, leaders don’t settle for one option. Most establish extensive audience measurement with Automatic Content Recognition (ACR) technologies. 

ACR matches individual objects in a video with database records to identify and recognize streaming content. The technology includes either or both video pixel detection (video fingerprinting) and audio capture (acoustic fingerprinting).

ACR-supported devices (smart TVs, smartphones, and tablets) allow ad networks to capture these data points: 

• Platform type – linear, CTV, MVPD, or another VOD service 
• Geo-location data 
• IP address 
• Demographics data 
• Viewing behaviors – average watch time, ad competition rates, channel surfing parameters, etc. 

Tech-wise, ACR algorithms generate library-side fingerprints for the publisher’s media. Fingerprints are designed to compare sample video/audio content against references in the publisher’s database to identify the played content. When a viewer browses content via an ACR device, they generate extra fingerprints, which then get matched to stored records. 

Based on matches, AdTech platforms access the above data for targeting, measurement, and attribution. Next, ACR data can be cross-validated with passive or digital metering for even higher accuracy. 

iSpot audience measurement with ACR 

iSpot has developed a robust cross-channel TV measurement tech suite for detecting ACR-sourced ad impressions across 83 million smart TVs and set-top boxes. Following its 2023 acquisition of 605, the platform combines smart TV data from VIZIO and LG with set-top box data from 16.6 million homes.

The platform relies on intelligent algorithms for matching impression counts against set-top box data and a person-level panel for extra precision, with direct integrations with over 400 streaming publishers. Separately, ad impressions are verified manually by a team of editors.

Such a comprehensive TV ad measurement stack, bolstered by four acquisitions since 2021, has made iSpot a leading challenger to Nielsen. Its publishing partners include NBCUniversal (which certified iSpot as a cross-platform currency vendor), Warner Bros. Discovery, Paramount, and Roku, among others. On the AdTech side, iSpot has secured deals with The Trade Desk, Google, and an exclusive data partnership with TVision.

Figure out how to best report on CTV ad performance

Brands can track connected TV ads using standard performance metrics like ad viewability, quartile rates, and completion rates. However, these don’t always provide an accurate picture. 

Ad verification firm DoubleVerify found that one in four CTV platforms continued playing content, including recorded ad impressions, after the TV set was turned off. Ouch, this better get fixed, and it likely will be. 

In June 2022, GroupM launched an initiative to co-create a streamlined measurement framework and best practices for verifying that ads only get served when CTV screens are on. A joint study with iSpot found that 8-10% of streaming impressions play when the TV is shut off. Companies including Disney, LG Ads Solutions, NBCUniversal, Paramount, VIZIO, Warner Bros. Discovery, and Fox/Tubi committed to the effort. 

The initiative has since evolved, with NBCUniversal and GroupM conducting successful tests in 2024 using IAB Tech Lab’s Ad Creative ID Framework (ACIF) for cross-platform ad tracking.

DoubleVerify has continued to expand its MRC-accredited CTV measurement capabilities. Its Fully On-Screen certification, first accredited in 2021, ensures ads are only displayed when TV screens are on. In April 2024, DV earned additional MRC accreditation for Video Viewable Impressions in CTV, which is significant given that DV’s research shows over one-third of CTV impressions serve into environments where ads fire when the TV is off, contributing to an estimated $1 billion in wasted ad spend annually.

IAB also recommends using the cost-per-completed viewable view (CPCVV) metric since it’s the most efficient and value-driven option. 

Provide tools to track brand lift and incremental reach 

Most advertisers choose CTV to improve ToFU metrics like brand awareness and consideration. Also, they want to understand how many unique audiences OTT video campaigns engage on top of linear TV campaigns. 

Respectively, buyers want to see brand lift and incremental reach stats in their dashboards. In CTV/OTT advertising platform development, you have several ways to deliver these stats.

Brand lift tracking options:

• Partner with CTV/OTT providers and/or third-party measurement companies to access intel.
• Employ statistical modeling methods to estimate CTV ad exposure. 
• Augment extrapolated data with passive exposure tracking panels, such as mobile metering and fingerprinting technologies.
• Issue in-device surveys to capture viewers’ sentiment towards promoted brands. 

Incremental reach tracking

• Use ACR technology (audio or acoustic fingerprinting) to identify consumed content and viewing patterns. 
• Add a passive metering device to capture audio watermarks for higher precision. 
• Combine ACR data with device graphs to better distinguish between users who saw linear vs. OTT campaigns (and vice versa). This tech combo can also help retarget exposed users with a sequential campaign across channels, plus re-optimize display frequency. 

Consider ML-based contextual targeting as an add-on 

ACR is a firmware-based solution. ML-based contextual targeting is a conceptually similar solution, but on a software level. This option might be better suited for AdTech companies that don’t want to source ACR data from multiple CTV platforms. 

Apart from monitoring user behaviors similar to ACR, ML-based contextual targeting systems can:

• Forecast advertising inventory volumes across networks 
• Model accurate campaign performance predictions
• Facilitate audience segmentation and data-driven audience modeling 
• Promote better CTV ad fraud detection and prevention 
• Improve user/device identification and ad measurement tracking 

Combined, these qualities make ML-based contextual targeting a competitive add-on for your ad network. 

Integrate a third-party CTV ad measurement SDK

At the end of the day, brands want guarantees. Many CTV platforms have already voiced their support for OM SDK:  

• Apple TV
• Amazon Fire 
• Android TV (Google TV) 

What about the remaining options like Roku, Samsung Tizen, LG Web OS, and others? If you work with those providers, you’ll have to build a custom SDK for integrating third-party measurement partners. You can turn to professional tech consultants like Xenoss to build a custom SDK for integration and resolve other challenges of the CTV/OTT advertising platform development.

Final thoughts 

Connected TV advertising is still a “Wild West” for AdTech providers. Some chose to go “cowboy style” and accelerate their entry into this environment without having CTV ad measurement and attribution tools. This tactic might have worked a couple of years back, but in today’s swiftly maturing CTV landscape, vendors that cannot send a wealth of data down the bid stream will soon turn obsolete. 

As CTV platforms continue to compete with one another for ad dollars, smarter AdTech players can focus on developing better CTV measurement solutions to fit into this nascent ecosystem.  

Want to be at the vanguard of CTV ad measurement? Xenoss can help you get there with our in-depth AdTech market expertise and technical know-how. Contact us to discuss your project.

The post CTV measurement: AdTech stack for the fragmented market appeared first on Xenoss - AI and Data Software Development Company.

What draws brands to programmatic DOOH? 

In short, advertisers are interested in high-precision targeting and clear-cut ROI for a broadcast reach of digital out-of-home. For years, teams struggled to measure the effectiveness of out-of-home ads and attribute positive lifts in key metrics to such campaigns. 

Programmatic DOOH solutions solve this problem by bringing the advertising experience closer to audience-driven buying of digital ads.

In this post, we unpack:

• DOOH meaning for the advertising industry (and the big hopes behind it!)
• How programmatic DOOH works and what features DOOH systems have
• Why now is the right time to develop programmatic DOOH products
• Unique tech challenges AdTechs have to account for
• Latest market trends and developments in the DOOH industry

What is DOOH?

Digital out-of-home advertising (DOOH) combines hardware and software technologies for displaying dynamic ads in public spaces.

Think your average billboard, but on an HD digital screen and updated in real-time, based on real-world conditions such as weather or audience demographics.

Digital OOH can be highly contextual and creative. You can run short video reels, create interactive consumer experiences, or personalize the ad based on current events — sports scores, traffic conditions, or even passing planes. DOOH ads can also be configured to generate collect customer data, measure viewer sentiment regarding your brand, or generate leads on the spot.

This translates to higher view rates, better brand recall, and follow-up actions.

I think marketers see digital OOH as a great alternative to reach people in the same hyper-relevant way as with digital, but in a channel that can’t be skipped or blocked.

Lauren Sak, Senior Marketing Director at Intersection

Due to the novelty of DOOH ads, consumers are more likely to engage with them.

In fact, 76% of DOOH viewers take action (watching videos, visiting promoted stores or restaurants) after interacting with the digital billboard.  

Finally, DOOH can be programmatic. Innovative digital out-of-home advertising companies like Lamar and Broadsign allow brands to purchase out-of-home ads at selected locations and run them at fixed times. New market entrants are sizing up custom DOOH platform development, too.

Demand for programmatic DOOH is also on the rise. 32% of US advertisers rely on a combination of programmatic and manual buying, and 28% of surveyed respondents rely exclusively on programmatic campaigns.

Other benefits of programmatic DOOH include:

• Ability to run trigger-based buying campaigns
• Innovative ways to target consumers
• Higher brand recall and awareness
• A wider audience reach a lower cost

How programmatic DOOH works: Technical architecture overview

DOOH systems have two key elements:

• Connected hardware, often equipped with cameras and sensors.
• Software backend, featuring a combination of modules for dynamic ad displays, data capture, and subsequent data analysis.

A simple DOOH system can have these components:

Such a device can be connected to a supply-side platform (SSP). The DOOH SSP, in turn, proposes the available inventory to a demand-side platform (DSP), where advertisers can place real-time bids on available inventory. Essentially, you get the same programmatic ad buying experience as for digital ads — but you purchase placements in the physical world.

The latest versions of DOOH devices also come with extra capabilities.

Environment recognition

A DOOH device can be equipped with multiple sensors:

• Temperature gauges
• Accelerometers
• Air quality sensors
• Motion sensors

These sensors can be used to create contextual ads and trigger-based buying campaigns, which fuse physical and digital realms.

For instance, as part of the “Magic of Flying” campaign, British Airways installed a digital billboard in London, equipped with an ADSB antenna. Each time a BA plane flew over the area, the billboard automatically displayed an ad, synchronized to the flight path of the plane. Such creative dynamic content significantly enhanced viewer engagement with the ad and improved brand recall.

Measuring foot traffic

Lack of measurability often deters advertisers from OOH. Programmatic DOOH changes that. You can know how many people had the potential to view your ad. You can also analyze how popular each area is to estimate the possible ad impression count.

There are different methods for measuring foot traffic next to DOOH devices:

• Smartphone counts
• Infrared (IR) sensor counts
• Using pressure sensors
• By combining sensing technology with computer vision

CityTraffic, a creation of The Netherlands company Bureau RMC, conducted foot traffic measurements in some 620 European cities, across 600 shopping streets and 110 events with high precision and with all privacy considerations.

They use a combination of the stereoscopy-based scanner, infrared sensors, mobile device MAC addresses sensor, and a mobility viewer device equipped with computer vision. This combo allows them to measure unique footfall at different locations. Many DOOH inventory providers rely on a similar approach for foot traffic measurement.

Motion and gesture detection

The latest DOOH systems include a camera connected to a computer vision system. Such a setup lets you collect non-personally identifiable audience data such as age, gender, or facial expression attributes. You can also use motion detection systems to active ad showings and deliver an immersive brand experience.

British energy company E.ON used Ocean Outdoor’s network of digital out-of-home screens in Manchester and Birmingham to create a socially conscious “Let’s clean the air campaign.”

Each screen live-streamed the person within the detection range and the amount of pollution they were breathing in at the moment (using real-time data). Messaging changed depending on the pollution levels. The campaign attracted over 2,500 U.K. residents in one weekend and drove a positive lift in brand perception.

Geo-targeting and retargeting capabilities

Sensor-based DOOH systems can also process location data for retargeting. For example, you can track the number of Bluetooth-enabled devices in the area or tag users by their phone’s MAC address. Then supply this data to advertisers for optimized targeting.

Hivestack — a full-stack programmatic digital out-of-home platform — helped Mazda create a high-precision geo campaign built around custom audiences. Using available geofencing and mobile IDs data collected by DOOH devices, Hivestack pointed Mazda towards the optimal DOOH locations for running their ads. Then the Mazda team programmatically bid on open RTB ad impressions from DOOH SSPs, buying inventory that meets their custom audience criteria.

As a result of this campaign Mazda enjoyed a:

• 21% lift in aided ad recall
• 24% lift in brand perception
• 3% lift in brand behavior

Interactive elements

DOOH systems are more than “big screens.” They have connected devices with computing and data processing capabilities. Therefore, advertisers can easily integrate third-party data into their campaigns to make them more interactive and personalized.

DOOH software platforms can process:

• Point of sale data
• Social media feeds
• Weather data
• Sports scores
• Pollution levels
• Traffic data

…and other third-party insights, obtained from data brokers.

In a recent DOOH campaign, Skoda used location and live traffic data to show passersby how long it would take them to drive to one of the U.K.’s beautiful holiday destinations. For an automotive company, that was a refreshing take on advertising. Instead of promoting the technical characteristics of their new SUV, Skoda chose to focus on the “lifestyle aspect” of car ownership. And that landed well with their target audience — families.

Interactivity also lends extra engagement to DOOH ads. An Ultraleap study found that compared to static DOOH, dynamic DOOH ads have 21% longer dwell time and result in 2X more conversions. Also, viewers spend 50% more time viewing the ad, and they are 52% more effective in increasing brand awareness.

Why invest in the development of programmatic DOOH products

Brands are intrigued with the new omnichannel customer targeting possibilities of DOOH.

According to an Alfi study, 96% of senior advertising executives believe DOOH data can improve campaign creativity and allow brands to leverage even more granular targeting.

Not only are brands now able to utilize the same audience data across channels for targeting and activation, but the increased flexibility means that mid-campaign optimization can now be applied to DOOH. For example, the best locations for driving in-store traffic or mobile downloads can be upweighted at the click of a button, and advertisers can see the impact of each media within the campaign mix and adjust accordingly.

Helen Miall, CMO of  VIOOH 

Here are five solid reasons to add programmatic DOOH to your AdTech software development roadmap. Larger advertisers are looking for high-precision targeting, transparent reporting,  and creative campaign styles. Programmatic DOOH ticks all of these boxes — and lets you optimize your operating margins too.

Advanced attribution

Programmatic DOOH lets you match device-collected data with audience insights from third-party attribution vendors to provide more precise targeting. A comprehensive data ecosystem allows advertisers to run high-performance omnichannel campaigns with DOOH in the mix.

Pepsi Max recently hosted a series of tasting challenges in malls. To retarget those prospects, they logged a unique ID of each participant using beacon technology. Then when one of the tasters entered a mall, Pepsi automatically triggered programmatic DOOH ads on screens. Clever and effective.

Data-rich inventory

DOOH can provide media buyers with rich data on each inventory asset — from average foot traffic to average viewability or ad interaction rates. This makes inventory more appealing to brands — and more profitable for DOOH system owners.

With DOOH, advertisers can purchase ad units in locations most popular with their target audience, perform advanced segmentation, or run sequential ad campaigns across channels. For example, target transit passengers with mobile ads first. Then retarget them with a related ad on a digital screen at their final destination.

Nestlé Purina, for example, leveraged data from Otto Retail to target the audience of cat owners. Based on this first-party data, they’ve selected optimal DOOH ad placements and the best time to display them. Simultaneously, they targeted this audience via online radio channels. The campaign was executed programmatically, which allowed Nestle to boost impression count by 13% without increasing the budget.

Predictive modeling

AdOps teams can further increase the precision with which DOOH captures customers at the point of maximum possible engagement once they embed predictive analytics into the DOOH stack.

Identifying engagement patterns helps media buyers estimate:

• Which locations will yield higher engagement
• What time is optimal for capturing more ready-to-buy passerbyers
• Which ad spend should the team allocate to the campaign

For DOOH vendors, expanding their offerings with predictive analytics helps retain partners and scale their impact in the client’s ad spend.

For example, after a successful DOOH campaign for Anytime Fitness, Vistar Media successfully used collected data to plan the second flight that captured a higher number of relevant venues and generated a 15% increase in sign-up intent compared to the first campaign.

Better ad experience

“Banner blindness” and high usage of ad-blocking software render digital ads less effective. Likewise, many standard digital ad formats don’t allow creating immersive viewing experiences (except for in-game advertising and native ad placements).

DOOH ads, on the other hand, can bridge the physical and digital worlds. The ad creative can be updated dynamically to be more personalized and memorable. DOOH can tie ad messaging to real-time events — weather conditions, the latest game scores, or the number of cars in the area.

For instance, ​​Sea-Doo managed to get an 80% lift in purchase intent after running a weather-based DOOH campaign. Using Foursquare’s audience and POI targeting, the watercraft seller ran ads across several key US locations with dynamic messaging, suggesting that a cloudy day shouldn’t deter you from taking a boat ride.

Easier ad rotation

Unlike standard OOH, you don’t need to change any marketing collateral once the campaign period expires. Programmatic execution lets you rapidly switch between campaigns moments after the impressions were delivered, lowering your management costs and increasing profit margins.

Better pricing dynamics

Instead of entering fixed-price agreements with advertisers, you can run real-time auctions based on OpenRTB standards. Brands can bid on available DOOH inventory and snatch the best deals for the lowest price. Or settle for the next-best option.

This allows you to adjust pricing to the current supply and demand dynamically. At the same time, DOOH owners will get higher fill rates. You can also set up your AdTech platform to support programmatic guaranteed deals or private marketplace advertising deals to retain loyal brands.

Synthetic audiences

Now that the regulations around collecting deterministic user-level data are getting tighter, brands and AdTech vendors are increasingly tapping into synthetic data capabilities.

With generative AI and predictive analytics, DOOH vendors can build audience segments that match the age, income, interests, habits, and movement patterns of real-world audiences. Media buying teams can use their understanding of this traffic to plan campaigns, collect data, and onboard new screens more effectively.

MOVE, an Australia-based DOOH audience measurement company, is already tapping synthetic audiences to help brands better understand local consumers. Its AI-augmented dataset accurately represents 2 million Australians over 14 years old, which is approximately 10% of the country’s population. Based on audience data, MOVE helps brands simulate the moving patterns of target customers and build detailed demographic profiles.

The company’s data modeling technologies reliably support DOOH market leaders, including JCDecaux, Metrospance Outdoor Advertising, APN Outdoor, QMS, and many others.

Hence, for an AdTech vendor, rolling out proprietary synthetic data capabilities can become a powerful differentiation point that helps both attract brand demand and build industry partnerships.

Tech challenges of DOOH

DOOH advertising solutions development requires knowledge of both hardware and software components of the ecosystem. Hardware market fragmentation alone can pose major roadblocks.

Since it’s a new channel, DOOH also has fewer technological standards for programmatic ad serving. At the same time, you also must account for new data types and unique creative formats.

But these shouldn’t phase you, especially if you are working with an experienced AdTech development partner.

Measuring ad viewability

To deliver effective measurement, DOOH devices have to be equipped with HD cameras and robust computer vision systems (which are complex to develop in the first place). This tech combo ensures proper rendering of the environment and ad viewability measurement.

But here’s where things get tricky: You also have physical device constraints. DOOH ads may not be easily visible from every angle. Likewise, it would be best if you minimized passerby double-counts.

OAAA attempts to address DOOH ad measurability issues with set guidelines and best practices. To accurately calculate viewability, they suggest factoring in:

• Distance between the user and DOOH display (varies by venue and device type)
• Latitude and longitude coordinates of the screen
• Average dwell time, based on the consumer’s proximity to the screen
• Cardinal direction that a screen faces

To deliver accurate reporting to buyers, you must capture and analyze a host of new input variables for different types of inventory. The Frankfurt airport employs an innovative DOOH measurement solution, designed by leading global specialists from JCDecaux and Veltys specifically for airports worldwide:

Nonstandard creative formats

DOOH screens come in different shapes and sizes. If you plan to add a DOOH asset to your inventory, you should verify that it can serve ads in adaptive HTML5 format. HTML5 allows advertisers to quickly adapt their content from other channels (mobile, web) to DOOH campaigns.

If you plan to sell video DOOH ads, pay attention to content length. The creative has to be short, 5-10 seconds long. Serving video DOOH ads of widely different lengths can mess up your broadcast scheduling. Some DOOH devices can incorrectly display too short or too long playouts.

The Digital Media Institute (DMI) released recommended specs for video and visual DOOH campaigns. You can (and should) make these part of your requirements for content.

Real-time data

The ability to tailor creatives to real-time traffic, weather, or sensor data is a major part of the DOOH appeal, and it is a non-trivial technical challenge.

Brands and DOOH companies also have to carefully navigate privacy challenges around real-time data collection.

30Seconds Group, a UK-based digital billboard company, faced backlash for using face-tracking cameras to monitor how apartment block residents respond to ads. One of such residents voiced concerns about an AdTech company spying on him in a comment for The Guardian.

RMG says I’m not being spied on, but there are cameras in the devices; you can see them. Even if it was at zero cost to residents, I would still fight these tooth and nail, nobody wants to be spied on by 6ft garbage adverts in their own building.

To avoid public scrutiny, DOOH vendors need to look for alternative data collection tools – live feeds, mobile SDK location data, on-site sensors, QR codes, or Bluetooth.

But, even with a pool of reliable data sources, building a data pipeline that will both display a personalized creative in under 100 milliseconds and scale to serve millions of impressions (this is the scale at which market leaders like Vistar operate) requires strong in-house data engineering capabilities.

When committing to building a pDOOH platform, make sure to select vendors with a proven track record in four areas.

• Designing a pipeline that supports both batch and streaming processing
• Enforcing data quality gates to prevent false or irrelevant data from triggering ad display
• Setting up low-latency integrations with other AdTech intermediaries (DSPs, SSPs, CDPs, and CMPs)
• Building an error-proof creative optimization and content delivery engine.

Working with a team that understands the nuances of low-latency, high-scale architecture of pDOOH solutions will help protect data security and avoid display errors that dissipate brands’ ad spend.

Hardware limitations

DOOH systems have become more advanced. But there are still some inherent hardware limitations. By design, not all systems allow establishing proper programmatic ad serving. You will need to create a prescreening mechanism for media owners to accept only suitable suppliers to your ecosystem.

For example, the DOOH device must have sufficient CPU, GPU, RAM, and storage to display HD content correctly. Also, it should support video codecs your platform can process and have all the needed connectivity options — WiFi, Bluetooth, 4G/5G, etc.

The digital screens market is highly fragmented, so it’s best to decide on some limitations instead of trying to optimize your platform for every type of device.

Proper targeting and attribution

Interactive DOOH systems process data in multiple formats — camera video, sensor data, Bluetooth-enabled devices capture, and data from third-party providers. These data points are necessary for high-precision targeting and attribution.

To ensure proper tracking and analytics, you need to develop a secure, high-load data management platform. Any glitches or inconsistencies can undermine the credibility of your DOOH measurement and reporting. Rapid data matching and processing are also crucial to avoiding lags in ad delivery and targeting efficiency.

Dmitry Sverdlik, CEO at Xenoss

Integrating DOOH inventory into programmatic platforms

The advertising industry has yet to rule on clear-cut standards around movement, ad play, and venue data, which are necessary to establish ad viewability.

The data variables themselves can tell conflicting stories. For example, the direction that the outdoor screen is facing can help validate the travel direction of a mobile device. But it’s a less relevant metric for indoor displays as passersby can pedal back to look at the ad. But not all DOOH hardware can provide this information.

When it comes to programmatic DOOH buying, there’s also no consensus on which data points DSPs and SSPs should exchange. Many platforms fail to factor in the unique characteristics of place-based advertising, such as:

• One-to-many vs. one-to-one impression delivery
• Extra latency in ad delivery for larger creatives
• Pixels for video tracking won’t work as an accurate measurement

The Digital Place-Based Advertising Association (DPAA) developed a framework for programmatic DOOH based on the OpenRTB 2.5 protocols. But with adjustments, accounting for the unique requirements of DOOH ads.

Privacy considerations

DOOH devices can collect more user data — from location to demographics. But requirements around user consent for such data collection vary by country.

Consumer privacy regulations such as GDPR and CCPA set rigid standards for collecting, storing, processing, and disclosing customer information in the EU and the US. Because of these, DOOH providers cannot transfer live video from camera systems — and process only text-based attributes. That’s called anonymous video analytics.

Computer vision-based DOOH devices can only perform facial detection, not facial recognition. The device can scan the consumer’s expression, age, or gender but not directly ID them based on unique facial features. In fact, 54% of US consumers are opposed to advertisers using facial detection technology to measure their reactions to public ad displays.

But the sentiment is different in the East. China, for example, has more relaxed privacy regulations. Back in 2015, China’s postal service did a multi-city DOOH campaign, using displays that tracked the viewers’ eye movements and dwell time of each glance while also factoring in the “biometric signature of each individual.” The country also largely normalized the use of facial recognition technology in “cashless” stores and hotels where customers can check out using their faces.

Ubiquitous connectivity, a wide network of CCTVs, and the newest digital screen models have made China a booming DOOH market with advanced targeting options.

The state of DOOH market

In the US, DOOH ad spending is projected to reach $2.87 billion by 2027.

Much of the industry growth will come from a rapid programmatic DOOH expansion with RTB opportunities now becoming available via mainstream DSPs.

In 2018, JCDecaux — a global leader in outdoor advertising – launched a programmatic out-of-home trading platform (VIOOH). Since then, they’ve been adding thousands of new DOOH devices to their global network. VIOOH recently added Frankfurt Airport to its media portfolio. The fourth busiest airport in Europe implemented a DOOH system across 23km2 of its area.

Through VIOOH, advertisers can now access over 800 panels of Frankfurt Airport in 34 DSP via PMP. JCDecaux (VIOOH’s parent company) currently provides airports DOOH inventory programmatically across the US, EMEA, Asia, and Australia.

AdTech startups are also expanding into programmatic DOOH with the help of venture capital. In 2021, Place Exchange, a DOOH SSP platform, closed a $20 million Series A round. Vistar Media, an end-to-end programmatic platform, secured $30 million in a Series B the same year.

Overall, the DOOH market is merely entering the growth stage. Over the next two years, 95% of advertising executives expect the DOOH market to grow significantly and surpass $50-$55 billion by 2026.

Programmatic DOOH marketplaces

Entering the DOOH market now can still give you the “first mover” advantage and the ability to secure contracts with large brands before they select an alternative provider.

But you must move fast, as other AdTech players are already carving their initials in the markets.

DOOH DSPs

The following companies specialize exclusively in DOOH inventory or have extensive access to it:

DOOH SSPs

The following companies allow digital out-of-home media owners to list their inventory or leverage their own inventory:

Final thoughts

Programmatic DOOH is an uncharted new territory to conquer. It comes with a host of obstacles, mainly around data processing, ad viewability measurement, and low-latency ad creative processing. But those that resolve these issues will be well-positioned for upcoming growth.

Advertisers are looking to scale beyond private marketplace deals and programmatic guaranteed ad placements. Many also want to run dynamic, data-rich campaigns in locations frequented by their ideal targets. But few players deliver that type of end-to-end buying experience. Your company can fill in this gap.

Xenoss can help you with DOOH integration to your AdTech platform or develop a new DOOH DSP/SSP platform. Contact us to discuss your project.

The post Digital Out-Of-Home advertising: Benefits and challenges of implementing programmatic DOOH appeared first on Xenoss - AI and Data Software Development Company.

As spending, viewing hours, and advertiser budgets shift toward CTV, publishers need to choose the right monetization model.

The two dominant CTV revenue paths are:

•  SVOD (subscription video on demand)
• AVOD (ad-supported video on demand). 

Each offers massive scale opportunities but comes with operational challenges, retention concerns, and infrastructure requirements. 

SVOD continues to expand globally, with households maintaining an average of four paid subscriptions. Markets like MENA are projected to reach $1.5B in streaming revenue by the end of this year. 

AVOD is accelerating, too. 90% of European marketers plan to increase AVOD/FAST spending in 2025. Nearly 80% of consumers say they will accept ads if the content is free. 

However, neither model is flawless.

SVOD faces rising acquisition friction, declining perceived value, and churn rates reaching 50% among Gen Z and millennials. AVOD deals with fragmentation, measurement gaps, and CTV fraud. 

As publishers aim to balance predictable subscription revenue with scalable ad revenue, the hybrid model is becoming the new standard in streaming. 

Netflix, Disney+, and Prime Video have fully integrated AVOD into their streaming experiences, expanding both revenue and user bases.

In this article, we’ll examine how publishers can blend SVOD, AVOD, and hybrid monetization strategies, compare the costs and benefits of both, and offer an actionable roadmap publishers can follow to monetize streaming services. 

Why SVOD gives CTV publishers a strategic advantage

For web publishers, a shift to paywalls and subscriptions came with considerable friction. Industry surveys show that only 17% of readers pay for news media, and 83% simply move on to a free source covering the topic when they hit a paywall. 

Despite the headwinds, news publishers are committed to subscriptions because the upside is much higher. Even though web publishers report online traffic decline since paywall adoption, 76% still saw higher reader revenue, and the average ARPU rose from $24 to $29. 

Streaming services have it easier because subscription-based video-on-demand (SVOD) has been the default business model. 

In 2025, an average American household is comfortable paying $70/month for streaming services, so SVOD publishers don’t face the same attrition as news media do. 

In fact, until a publisher has a wide enough reach and content library to explore ad-supported monetization, SVOD should be the default monetization playbook for a few reasons. 

1. SVOD creates stable, predictable revenue

CTV ad spend is growing, but the market is still volatile and relies heavily on macroeconomic trends. 

Linear TV is a clear example of how relying purely on ad-based monetization makes publishers more vulnerable to shifts in ad spend. In December 2025, German broadcaster RTL had to lay off 600 staff members due to a dip in ad revenue and a lack of alternative, reliable income sources.  

On the other hand, while both Disney and Paramount reported a decline in ad revenue in Q3 2025, both publishers run a SVOD business model, which cushioned the impact of a weaker ad quarter. 

Relying on monthly subscription fees on the outset of launching a streaming service helps create a brand-loyal community of viewers that fuels recurring revenue. 

Publishers can funnel SVOD returns into expanding the content library, engineering infrastructure, and supply chains on a stable basis before they are ready to layer AVOD as an additional revenue stream.

2. SVOD is the strongest source of first-party data

A SVOD offering encourages publishers to build direct connections with their audiences. These relationships are account-based and authenticated, with viewers logging in, sharing emails and payment details, and building long-term viewing histories tied to a persistent ID. 

Over time, SVOD publishers can build a long trail of data on viewing habits, session length, devices, and genre affinity. 

Considering that AdTech has been on the edge about cookie deprecation for the last three years, having a robust first-party data library as a backup plan differentiates SVOD publishers from media that rely solely on third-party trackers. 

3. SVOD still makes room for branded deals and advertising integrations

Subscription-only platforms typically avoid interruptive advertising, but they can still monetize brand partnerships through:

• product placement
• branded content
• native integrations
• co-marketing campaigns

These formats allow publishers to capture high-value brand deals without sacrificing user experience, requiring an in-house AdTech stack or sharing ad revenue (in some cases, up to 50%) with advertising partners. 

A well-known example is the Eggo waffles product placement in the Netflix show “Stranger Things”, which brought a 14% sales increase in 2017 and a 9.4% sales uplift in 2018. 

The limitations and risks of SVOD 

The SVOD industry faces a mounting credibility crisis as consumers increasingly question whether their subscriptions deliver real value. 

While 53% of consumers rely on streaming services as their primary paid entertainment source, satisfaction is plummeting. 

Now that more SVOD platforms are hitting the market, an average household in the US has to maintain four active streaming services. Having to pay a separate monthly subscription for each of those makes one in two viewers feel like they are spending too much on CTV content. 

As a result, SVOD publishers are now facing a harder time acquiring new subscribers and retaining their audiences. 

1. Growing customer acquisition costs

In the last three years, SVOD publishers have had a harder time retaining viewers whose attention is dispersed on short-form social media content. 

With high-quality video generation models like Sora and Nano Banana, the sheer volume of available video content is growing exponentially, making it harder to cut through the noise. 

A Deloitte survey on digital media trends noted that SVOD publishers are falling behind on personalization expectations of younger audiences and are losing viewers to social media, where algorithmic recommendations reflect user interests more accurately. 

To continue acquiring new subscribers, SVOD streaming services invest more in:

• sophisticated recommendation engines
• social media campaigns promoting new releases
• bundles, discounts, or extended free trials

These tactics help with acquisition but drive CAC higher every year.

2. Rising customer churn

Even when platforms succeed in attracting new subscribers, retaining them has become significantly harder.

Throughout 2025, subscriber churn has been rising. Deloitte reports that 40% of consumers have cancelled at least one paid streaming service every six months. 

The average churn rate among large SVOD publishers, Netflix, Hulu, and Disney+, is at 5.5%, a two-fold rise from 2.9% in 2019. 

Churned viewers are not lost forever. 24% of them resubscribe within six months. However, chasing these audiences requires publishers to keep running costly re-acquisition campaigns that erode the bottom line. 

The new monetization playbook: adding AVOD to an SVOD service

Viewers reaching the tipping point about the top price they are willing to pay for a streaming service is both a challenge for SVOD providers and an opportunity to explore adding a cheaper ad-supported video on-demand (AVOD) tier. 

Before Netflix rolled out ad-supported subscriptions, many industry analysts thought that ads would increase subscriber churn by making streaming more similar to linear TV, which it originally branched away from. 

However, according to industry signals, viewers no longer mind ads if they can save on subscriptions. 

• A Marketing Brew survey reported that 80% of consumers would accept ads if video content were completely free.
• Two-thirds of consumers surveyed by PwC say they’ll tolerate ads to lower subscription costs
• Ad acceptance is rising even among self-proclaimed ‘ad-haters’: 42% of them are now tolerant of ads in streaming platforms. 

Audiences primarily want access to more content at lower prices. As households juggle multiple subscriptions, adding AVOD tiers becomes an acceptable, even welcomed, trade-off.

Benefits of expanding SVOD capabilities with AVOD offerings

AdTech is ready for the growth of AVOD inventory

Besides becoming widely accepted by customers, in-streaming ads are heavily sought out by advertisers. 

68% marketers now view AVOD CTV channels as “must-buy” items, and demand will likely go up as the programmatic ecosystem for CTV matures. 

For now, this growth has been slow; most AdOps teams don’t have dedicated CTV advertising teams, and only 34% of the total CTV inventory is biddable. 

But the ecosystem is picking up pace. By early 2026, nearly half of CTV inventory is estimated to be biddable, and 75% of marketers plan to set up internal teams for CTV campaign management by the end of next year. 

Both advertiser interest and the rate at which tech capabilities grow are looking good for AVOD publishers. 

AVOD is a way to monetize the first-party data SVOD publishers collect

SVOD subscriptions generate high-quality, authenticated first-party data, but the data becomes significantly more valuable when publishers add AVOD capabilities. With both models in place, publishers can use viewer behavior, device usage, genre affinity, and title-level interaction data to create premium audience segments, higher CPMs, direct deals with global brands, and more accurate frequency and reach models. 

Real-life example: Disney+ centered its AVOD offering around high-quality first-party data

Disney Advertising has built a suite of high-value ad products on top of its first-party data to attract high-budget advertisers. The publisher’s Audience Graph and Disney Select tools aggregate streaming and other Disney touchpoints into more than 1,000–2,000 first-party behavioural and psychographic segments. 

Global advertisers like Chipotle, United Airlines, and T-Mobile tapped into Disney’s metadata and audience graph to insert ads in key emotional moments of Disney content and drive more user attention to their campaigns. 

Fueled by growing viewer acceptance, AdTech capabilities, and brand demand, AVOD is becoming the industry standard. Amazon, Disney, Netflix, Paramount, and many other leading streaming services are effectively running ad-supported monetization on top of monthly subscriptions. 

Why new publishers should not choose AVOD as their only monetization model

The rise of AVOD may tempt new entrants to skip SVOD entirely and launch as a free, ad-supported service. 

In our experience, this is a riskier strategy because building or buying an AdTech stack requires considerable upfront investment, both in engineering capabilities and internal sales teams. 

Need for a proprietary AdTech stack

To successfully support AVOD streaming, publishers have to run an ad server in a channel that’s still fragmented and lacks robust AdTech standards. 

To appeal to advertisers, publishers also need to circumvent inconsistent CTV measurement, disparate reporting, and a lack of data standardization with custom data pipelines, clean IDs, and cross-screen attribution. 

Building a competitive AdTech stack for AVOD will stretch time-to-market and require a considerably higher budget. For a new CTV market entrant, setting up a simple subscription pipeline first and investing all remaining funding into the content library makes more sense in the long term. 

Difficulty building engaged audiences

Major SVOD providers who have been experimenting with ad-supported streaming report that ad-supported users watch 22–23 minutes less per day than ad-free homes and churn faster than ad-free tier subscribers. 

Not having the support of a more engaged SVOD audience and scaling a streaming service built on less committed viewers exposes publishers to risks in viewership fluctuations and will likely make them less attractive to advertisers compared to services with combined SVOD and AVOD monetization. 

How streaming publishers can integrate both SVOD and AVOD monetization

The decision framework for adopting SVOD and AVOD comes from understanding their respective strengths and weaknesses in customer acquisition and content production costs, upfront investment in development, and margins. 

SVOD monetization is easier to build into a streaming platform than an AVOD stack, which is why all leading CTV publishers use it as the default model. It will help lay a strong financial foundation, more predictable retention curves, and a clear playbook for collecting first-party data. 

However, in a market where consumer price sensitivity keeps rising and subscription fatigue is accelerating, SVOD is no longer sustainable on its own. 

Introducing ad-supported monetization gives SVOD publishers the ability to cut subscription costs and improve user retention while maintaining positive margins and attracting new financial gains through ad revenue. 

Five-step framework for SVOD launch and AVOD transition

Drawing from our experience in building CTV solutions, we developed a five-step monetization roadmap that publishers can effectively combine SVOD and AVOD capabilities. 

Step 1: Launch with a tight, easy-to-understand subscription offer.

A focused content proposition, simple plans (1–3 tiers at most), and a smooth signup/billing experience across key devices.

Step 2: Instrument data from day one and build a clean first-party data flow. 

Require login for all subscribers and track viewing, engagement, churn, and acquisition channels in a unified data model. This first-party data becomes the backbone for later decisions on content, pricing, and, eventually, ad targeting.

Step 3: Stabilize unit economics before touching ads. 

Iterate on catalog, recommendations, UX, and pricing until you hit acceptable CAC payback, churn, and LTV/CAC ratios. Only once subscription revenue is predictable and reasonably profitable should you consider adding another monetization layer.

Step 4: Design an ad strategy that complements SVOD.  

Introduce an “ad-lite” or AVOD tier as a deliberate segmentation move. Lower price or free with registration, without degrading the value of your flagship ad-free plans. Clearly define which audiences each tier is for and how you’ll move users up the value ladder.

Step 5: Phase in AVOD infrastructure and optimise with SVOD data. 

Roll out SSAI, measurement, and IVT/fraud controls incrementally, starting with limited ad loads and a small set of trusted demand partners.  Use your rich SVOD first-party data to power targeting, frequency management, and content/ad load optimisation, so ads are a high-yield add-on rather than a structural dependency.

By following these implementation steps, CTV publishers can tap into fast-growing ad budgets without exposing themselves to ad-market whiplash. The services that win this decade will be the ones that continually rebalance the SVOD/AVOD  mix, using first-party data, unit economics, and viewer sentiment as their north stars. 

The post SVOD, AVOD, or a hybrid model: How streaming platforms can maximize CTV revenue appeared first on Xenoss - AI and Data Software Development Company.

Post-launch, the campaign metrics looked impressive: millions of completed video views, completion rates above 90%, and strong brand sentiment scores. 

The attribution dashboard showed their search campaigns and social media driving all the measurable results, while the expensive streaming advertising appeared to have contributed zero business value.

This CTV underperformance makes the C-suite question the entire CTV strategy. If it couldn’t prove its value, why allocate a significant budget to it?

Nielsen’s 2025 Annual Marketing Report shows 56% of marketers globally planning to increase CTV spending this year. Meanwhile, Analytic Partners’ ROI Genome research reveals that CTV averages just 7% of total advertising spend while delivering ROI that’s 30% higher than other marketing channels.

The math doesn’t work. A channel performing 30% better gets 7% of the budget, while underperforming channels capture the majority of marketing investments. Companies are missing revenue opportunities because they can’t prove which advertising channels drive sales.

Why CTV campaigns look successful but show zero ROI

Marketing departments are trying to measure streaming advertising with tools built for a fundamentally different media landscape.

Traditional TV measurement operated in a controlled environment. Nielsen panels provided audience estimates. Demographic assumptions predicted behavior patterns. Geographic regions determined the ad inventory availability.

CTV advertising operates across dozens of fragmented platforms that each control proprietary audience data. 

Roku manages viewing behavior for 80+ million households through its operating system. 

Samsung tracks smart TV usage across 200+ million devices globally. 

Amazon captures Prime Video engagement alongside shopping behavior for 230+ million customers. 

Netflix, Disney+, Paramount+, and regional streaming services each maintain separate measurement ecosystems.

This means brands trying to get a holistic view of performance are stuck piecing together partial data from each provider.

Even with strong creative and intelligent targeting, you can’t optimize or justify investment without a connected view.

The cross-device purchase journey problem

Customer purchases involve multiple devices and platforms over several weeks, but measurement systems only see individual touchpoints.

A working professional sees an ad for your brand while watching Netflix on their smart TV during evening entertainment. The ad addresses their specific pain point.

The next morning, they search for your product on mobile, visit several websites, read comparison articles, and bookmark solutions for later evaluation. At work, they conduct deeper research on their laptop, viewing product demos and reading customer reviews. The final purchase happens a lot later than the initial CTV exposure.

Standard attribution tools assign conversion credit to the last interaction before purchase, usually a direct website visit or email click. The original CTV impression that started the evaluation process gets zero credit. Reporting dashboards show the streaming campaign as ineffective while other channels appear successful. This misalignment distorts ROI calculations.

Analytic Partners found that 30% of paid search clicks are driven by other advertising, predominantly from video campaigns. Without CTV awareness building, significant portions of “successful” search conversions wouldn’t happen.

How to build cross-device identity resolution

Solving CTV attribution requires connecting viewing behavior across every device and platform customers use throughout their purchase journeys.

Start with deterministic data collection wherever possible. When viewers log into streaming services, they create concrete connections between devices and identity. These authentication events provide the foundation for household-level tracking that persists across viewing sessions.

Layer probabilistic modeling on top of deterministic connections to fill attribution gaps. Devices sharing IP addresses, geographic clustering patterns, and behavioral signal similarities often indicate household membership. Machine learning algorithms can analyze these patterns to predict household relationships with 75%+ accuracy when combined with observed behavioral data.

The most sophisticated approach involves integrating streaming exposure data directly with customer data platforms through real-time API connections. This requires data engineering infrastructure capable of processing streaming datasets at scale while maintaining privacy compliance and attribution accuracy.

Marketing Mix Modeling provides the analytical framework for measuring CTV’s incremental contribution across 60-90 day attribution windows. MMM analyzes statistical relationships between advertising inputs and business outcomes while controlling for external factors like seasonality, competitive activity, and economic conditions.

MMM studies reveal CTV delivers nearly twice the sales impact compared to other channels, with streaming investments proving twice as effective as traditional media and outperforming ROAS benchmarks by 22%.

CTV ROI problem #1: Frequency chaos wastes budgets while annoying customers

Consider the viewing patterns of heavy streaming users. They might encounter your ad campaign through Hulu during morning news, Netflix during lunch breaks, Amazon Prime during commute entertainment, and Disney+ during evening family time. Platform-specific frequency caps of 3x weekly become 12+ total exposures when multiplied across four streaming services.

LG Ad Solutions found an average CTV brand frequency of 7.3, but this statistic masks extreme distribution problems. Heavy streaming viewers experience frequency levels reaching 150+ weekly exposures, while casual users remain completely unexposed.

This creates the worst possible outcome: oversaturated audiences developing negative brand associations while high-value prospects never see your message. The problem intensifies during peak viewing when inventory costs spike and competition increases.

How to coordinate frequency across streaming platforms

Effective frequency management requires treating CTV advertising as a unified channel rather than separate platform campaigns.

Implement household-level exposure tracking that connects impression data from all streaming inventory sources. Use unified household identifiers that persist across platforms and devices to create comprehensive exposure histories. This provides real-time visibility into cumulative advertising pressure at the household level.

Establish global frequency limits that account for cross-platform exposure rather than individual service caps. Optimal CTV frequency ranges from 3-7 exposures per week for awareness campaigns and 5-10 for consideration and conversion objectives.

Deploy real-time bid modifications that redistribute budget from oversaturated households to unexposed audience segments. When exposure data indicates someone has seen your ad 6 times this week, reduce bids dramatically or exclude them from targeting entirely. Redirect that budget toward lookalike audiences with similar demographic profiles but lower historical exposure levels.

Even if your campaigns hit the right frequency targets, the story doesn’t end there. What happens next, how quickly your team responds to performance signals, can make or break ROI.

CTV ROI problem #2: Delayed optimization cycles burn budgets on poor performance

Most streaming platforms struggle with delayed optimization capabilities that prevent responsive campaign management. CTV ad-serving primarily functions as a ‘data out‘ system where measurement firms provide post-campaign reporting rather than enabling real-time optimization. 32% of CTV advertisers cite disparate reporting across multiple buys as their biggest challenge, making responsive optimization extremely difficult.

The measurement lag that kills campaigns

Marketing teams continue spending on underperforming ad variations for entire weeks before getting sufficient data to make informed optimization decisions. Most campaign management still operates on weekly optimization cycles, assuming that delayed decision-making won’t significantly impact results.

CTV campaigns frequently have concentrated flight schedules and limited budget windows that make every day of poor performance costly. A single week of running suboptimal creative can consume 25-30% of the total campaign budget before optimization interventions take effect.

The problem becomes more severe during competitive periods when CPMs fluctuate rapidly and audience availability changes throughout the day. Seasonal campaigns, product launches, and competitive response situations require real-time adjustments that traditional weekly reporting cycles simply cannot support.

Building real-time CTV performance monitoring

Solving optimization delays requires infrastructure that processes streaming performance data continuously and surfaces actionable insights within hours.

Connect directly to streaming platform APIs to extract impression, completion, and engagement data on hourly intervals rather than waiting for daily batch reports. Automated monitoring systems compare key metrics against predetermined benchmarks and trigger optimization workflows when performance patterns become clear.

Create unified dashboards that combine streaming performance with downstream conversion tracking to identify high-performing creative-audience combinations rapidly. For example, when specific ad variations consistently drive higher engagement rates, automated systems can shift budget allocation within 4-6 hours—the timeframe needed to accumulate statistically meaningful performance data.

Implement dynamic bidding adjustments based on real-time performance signals. If certain audience segments consistently show higher completion rates and stronger post-view engagement, increase bidding to capture additional inventory. When creative fatigue emerges through declining engagement metrics, automatically pause underperforming variations and redirect spend to fresh creative assets.

This would help to respond to performance changes before they consume significant budget, transforming CTV from a “set and forget” channel into a dynamically optimized performance driver.

Infrastructure that fixes CTV measurement problems

While most CTV platforms currently operate as “data out” systems with limited real-time optimization capabilities, addressing CTV’s ROI challenges requires a comprehensive measurement infrastructure that treats streaming advertising as a unified, data-driven channel.

Unified data collection architecture

Build data pipeline infrastructure that ingests streaming exposure events, device-level engagement signals, and downstream conversion outcomes into centralized systems designed for advanced processing. Use Apache Kafka for streaming data ingestion and Apache Spark for real-time analysis to handle the volume and velocity requirements of enterprise CTV measurement.

Implement custom API integrations that connect streaming platform data with customer data platforms, enabling automatic matching of ad exposures to known customer profiles and conversion events. Unlike social platforms that enable immediate optimization through conversion APIs, CTV requires this integration layer to provide the technical foundation for accurate cross-device attribution and long-term customer value analysis.

Machine learning attribution engines

Traditional rule-based attribution systems break down in the complex, multi-device reality of modern streaming consumption. Deploy machine learning models that analyze viewing patterns, engagement signals, device transitions, and time delays to predict conversion probability and assign attribution credit accurately.

Use these predictive models to power real-time optimization decisions: increasing bids for high-probability conversion audiences, pausing creative assets showing early fatigue signals, and redistributing budget from oversaturated households to unexposed prospects with similar conversion potential.

The most effective systems learn continuously from campaign performance data and improve attribution accuracy and optimization decisions over time. This creates compound improvements where better measurement enables better optimization, which generates superior performance data that further enhances measurement precision.

Implementation challenges and industry coordination

Building this infrastructure means tackling some real technical and industry roadblocks. Current CTV ad servers remain “blind” to business outcomes, optimizing only for impression delivery rather than conversions. Publishers naturally resist frameworks that might shift their ad revenue based on performance.

Processing data across dozens of fragmented streaming platforms while navigating privacy regulations like GDPR and CCPA adds complexity. But companies building these measurement capabilities now will control CTV advertising as the market shifts toward performance-based buying.

Case study: AI-driven dynamic creative optimization transforms CTV performance

A global consumer brand known for high-impact storytelling faced a common CTV challenge: their premium streaming inventory delivered strong brand recall, but generic creative messaging failed to maximize engagement across diverse audience segments.

The brand recognized that Connected TV’s non-skippable, full-screen format creates powerful storytelling opportunities. However, their static creative approach couldn’t adapt messaging for different demographics and behavioral segments in real time. Manual segmentation and basic creative rotation provided insufficient audience precision to match shifting viewer preferences.

Working with Xenoss data engineering specialists, the brand implemented an AI-driven Dynamic Creative Optimization system that personalizes video ads using real-time audience data while maintaining GDPR compliance and brand safety standards.

Technical implementation and business impact

The solution required infrastructure capable of handling device-level data and Automatic Content Recognition signals while making creative decisions fast enough for video ad serving without introducing delivery delays.

The system integrated diverse behavioral signals through privacy-compliant methods, using machine learning algorithms to match creative variations with audience segments automatically. Decision logic analyzed viewer patterns, demographic indicators, and contextual signals to select optimal creative elements for each impression.

Brand safety protocols ensured personalized creatives met guidelines for large-screen delivery across premium streaming inventory. The infrastructure processed creative selection decisions within milliseconds while maintaining creative quality standards and compliance requirements.

The AI-driven optimization delivered measurable performance improvements:

• 30% increase in view-through rates compared to static creative delivery
• Higher completion rates and improved audience retention during ad breaks
• Increased return on ad spend through efficient budget allocation to top-performing creative-audience combinations
• Stronger audience engagement while maintaining privacy regulation compliance

This implementation demonstrated how machine learning infrastructure can transform creative delivery in premium CTV environments, enabling brands to unlock richer storytelling capabilities while driving stronger business outcomes through data-driven personalization.

What’s coming next for CTV measurement and optimization

Amazon’s Prime Video advertising platform is building direct purchase attribution, connecting streaming ad exposure to same-day shopping behavior through their customer accounts. With Amazon’s closed-loop ecosystem linking Prime Video impressions to commerce behavior, this eliminates the multi-week attribution gaps that make CTV appear ineffective. Expect other retail-streaming combinations to follow: Walmart Connect with Paramount+, Target with Disney+ bundles.

Netflix’s advertising tier captured over 55% of Q4 2024 signups across the 12 countries where ads are available, creating household-level audience profiles that persist across devices and sessions. With Netflix’s ad-tier membership growing 30% quarter over quarter, this authenticated viewing data enables attribution accuracy that current probabilistic matching can’t achieve.

Samba TV and Kochava launched unified cross-platform measurement in July 2025, combining first-party viewership data from millions of connected televisions with advanced attribution analytics. This partnership delivers consistent performance insights across linear TV, connected TV, and digital platforms through a cohesive measurement pipeline that addresses always-on measurement at scale.

Privacy-compliant attribution methods using data clean rooms and durable identifiers to maintain targeting precision while meeting regulatory requirements. Organizations now gain sustainable advantages over competitors relying on diminishing tracking capabilities.

Automated optimization engines powered by machine learning that adjust bidding, creative rotation, and frequency capping in real-time based on conversion probability. Modern CTV platforms now deliver 45% higher conversion rates for exposed households compared to non-exposed audiences, demonstrating a measurable performance impact.

Integration with retail media networks creates new closed-loop attribution opportunities as streaming platforms launch advertising businesses and retailers develop CTV inventory. This convergence enables direct measurement of advertising exposure to purchase behavior, eliminating many current attribution challenges.

Bottom line

Your CTV campaigns might already be working. You just can’t see it.

The difference between streaming advertising success and failure isn’t creative, targeting, or budget size. Companies with working CTV measurement see results that others miss entirely. They make optimization decisions based on complete data.

Fix your measurement infrastructure, and you might discover your streaming campaigns have been driving business impact all along.

The post Why Connected TV ROI is underperforming and how to fix it appeared first on Xenoss - AI and Data Software Development Company.

For marketers, this creates a fundamental challenge. Traditional targeting methods are becoming obsolete. But here’s the counterintuitive reality: you’re not actually running out of data for personalization. Most teams are sitting on mountains of ultra-valuable, privacy-compliant information that customers willingly provide. Customer reviews reveal authentic preferences. Chat logs documenting real purchase intent. Voice recordings capture emotional nuance. Social posts showing genuine brand sentiment. Survey responses detailing exact needs.

The challenge lies in the nature of this data. It’s unstructured, messy, and non-tabular. Traditional marketing tools can’t process it effectively, which means most companies aren’t using it at all.

Multimodal AI changes this equation entirely. This technology can transform scattered customer conversations, heart-shaped emoji reactions, and voice note feedback into real-time, trust-based personalization that outperforms traditional tracking methods.

This guide reveals how forward-thinking marketing teams are combining multimodal AI with zero-party data strategies to create personalization that’s more accurate, more compliant, and more sustainable than anything possible with third-party tracking.

Zero-party data: Building the foundation of privacy-first marketing 

One of the most overlooked assets in every data stack is information customers give you on purpose, also known as zero-party data.

Forrester first coined the term zero-party data (ZPD) to describe “data that a consumer intentionally and proactively shares about herself” when interacting with your brand. This could be: 

• Preference center data: email frequency, product categories of interest, shipping addresses
• Purchase preferences: feedback on past orders, saved items, favorite delivery windows
• Personal context: content interests, hobbies, lifestyle indicators
• Self-identification cues: “plant mom”, “eco-conscious parent,” “anime fan”

ZPD represents the most accurate information you’ll get about a customer because it’s self-disclosed. Unlike first-party data, which is passively collected through user actions like page views or purchase history, or third-party data purchased from external providers, zero-party data starts with explicit consent rather than implied behavior. 

Zero-party data is explicit, not implied. Almost half of consumers are open to telling brands how they want to be seen and served if that leads to better experiences. This makes ZPD a powerful method for engaging and personalizing in a privacy-first world. 

Understanding the data spectrum: From third-party to zero-party

The distinction between different data types has become crucial for compliance and effectiveness. Third-party data comes from external sources without direct customer relationships, often through data brokers or tracking networks. Second-party data involves sharing between trusted partners with customer consent. First-party data captures behavioral signals from direct customer interactions. Zero-party data represents intentional customer sharing for better experiences.

Because zero-party data is shared willingly and proactively, its usage helps build trust. Three-quarters of consumers won’t buy from brands with poor data ethics, which typically involves pervasive tracking, ultra-targeted advertising, or a history of data leaks. Zero-party data addresses this by creating a permission-based loop where customers control the conversation.

When implemented correctly, multimodal marketing with zero-party data can unlock hyper-personalization without surveillance concerns. You can optimize content experiences across touchpoints in real-time based on self-disclosed customer preferences. You can feed user preferences into real-time bidding optimization models to prioritize higher-value segments for targeting. You can connect profile inputs to conversion optimization tools for more relevant offers.

Effective zero-party data collection methods

The most successful zero-party data collection happens through:

• Onboarding questionnaires that feel helpful rather than intrusive
• Email or app preference centers that give customers control
• Loyalty programs with profile enrichment opportunities
• Style or product discovery quizzes that provide immediate value
• Exit or post-purchase surveys timed for maximum engagement
• “Build your bundle” or configurator tools that capture preferences naturally
• Chatbot interactions that document self-reported intent

However, operationalizing zero-party data presents unique challenges. The datasets are smaller and harder to scale compared to automated tracking. Participation is voluntary, which means you must earn attention and trust before requesting information. When the exchange feels one-sided, users will opt out quickly.

To maximize zero-party data value, you need to combine it with first-party behavioral data obtained with proper consent. This combination enriches customer profiles, refines customer journey mapping approaches, and enables smarter automations across key touchpoints while maintaining privacy compliance.

Unstructured first-party data: The complementary asset hiding in plain sight

About 80% to 90% of all information produced today is unstructured. This includes customer service transcripts, support chat logs, social media comments, product reviews, voice recordings, and video interactions.

These assets often hold the context, rationale, and emotional nuance behind customer decisions. Structured data can tell you what happened. Unstructured data can tell you why it happened.

Yet unstructured data remains vastly underused due to persistent data silos and suboptimal IT infrastructure.

Leaders struggle to integrate structured and unstructured data because they don’t share common storage architectures. Structured data lives in SQL-based warehouses, while unstructured data resides in data lakes and object stores. Customer data integration becomes complex when these systems don’t communicate effectively.

Moreover, unstructured insights require advanced extraction, transformation, and modeling before they can be fed to standard marketing automation tools. This technical barrier has prevented most organizations from capitalizing on their unstructured data assets.

The untapped value in unstructured customer data

There’s significant upside to investing in streaming data pipelines and data lakehouse architecture that enables dual-type data processing.

 According to Deloitte, companies that use first-party data: 

• Get 8X higher ROAS from targeted ads 
• Drive 10% or higher  lift in sales 
• Are 44% more likely to beat revenue goals

When combined with zero-party data, these results can be sustained and amplified in the data protection era, thanks to new generation AI marketing tools that can process both structured preferences and unstructured behavioral signals.

Customer service transcripts reveal frustration patterns and feature requests that structured surveys miss. Social media interactions capture authentic brand sentiment and community conversations. Voice recordings contain emotional nuance, including tone, pace, and hesitation that text-based data loses completely. Product reviews include detailed use cases, comparison criteria, and decision-making factors. Chat logs document real-time purchase intent and consideration processes. Email conversations show relationship progression and engagement patterns over time.

The challenge has been extracting actionable insights from this messy, unformatted information at scale. Traditional analytics tools struggle with the variability and context-dependent nature of unstructured data, leaving valuable customer intelligence trapped in isolated systems.

Multimodal AI: The technology that unifies zero-party and unstructured data

Combined zero-party and first-party data offer richer context. One is volunteered, the other observed. But neither lives in the clean, structured rows that most MarTech tools can process effectively. Multimodal AI solves this integration challenge.

Multimodal AI models can process and synthesize different sensory modes, including text, audio, video, visuals, and all connected metadata within the same model. Unlike regular generative AI models primarily trained to excel at one job well, such as working with texts like GPT-4 or video like Stable Diffusion, multimodal algorithms aren’t limited to single input or output pairs.

They can be fed with different data modalities and prompted to produce any content type based on a comprehensive analysis across multiple signal sources.

The industry is moving rapidly in this direction. Gartner predicts that 40% of gen AI solutions will be multimodal by 2027, up from just 1% in 2023.

For marketing, the major advantage of multimodal AI is comprehensive customer behavior analysis. Such algorithms don’t treat customer preferences, reviews, voice notes, and chat transcripts as separate channels with context lost between interactions. The system reads them as one continuous conversation with cross-model validation, reducing ambiguities and knowledge gaps.

Multimodal AI bridges declared and behavioral intent

Multimodal AI creates a unified view of what customers say they want and how they actually behave and feel. This enables marketing teams to unify declared and behavioral intent, powering more human-like, context-aware engagement.

Multimodal AI marketing use cases

• Unified customer identities, based on a mix of quantitative and qualitative data 
• AI-powered personalization, driven by past customer support interactions or survey data 
• Cross-channel journey mapping by linking chat logs, app interactions, and purchase history
• Early churn prediction based on sentiment changes in emails, call transcripts, and product reviews
• Advanced sentiment analysis, based on social media posts, video reviews, and other UGC 
• Ad bidding strategy or creative selection, based on multimodal behavioral analytics
• Visual trend mapping from tagged images, reels, and unstructured community content

In short, multimodal AI unlocks real-time, nuance-driven personalization. By combining structured and unstructured signals, marketers can stop guessing and start creating campaigns that reflect what customers actually think, feel, and say. 

Implementation approach: Building the technical foundation for multimodal AI deployment 

Serving regular machine learning models requires a certain degree of data management and MLOps maturity.  You’ll need to fine-tune your current infrastructure to support faster data processing, seamless data sync, and effective infrastructure multi-GPU orchestration. On top, there’s also the cultural shifts to consider. 

Data architecture requirements

To run multimodal AI in production, you need the right data infrastructure behind the scenes. Legacy data processing infrastructure based on warehouses and rigid ETL pipelines won’t suffice. You’ll need storage infrastructure with flexible schemas to host video, voice, and clickstreams alongside structured tabular data.

The optimal solution is a data lakehouse architecture. This represents an open data management approach that combines the flexibility and scalability of data lakes with the reliability and governance controls of data warehouses. With a data lakehouse, you can run multimodal AI, predictive analytics, and other AI models on all data types, minimizing information asymmetry and blind spots.

Already, 55% of enterprises run more than half of their analytics on a data lakehouse, and 67% expect to reach the same benchmark within three years as part of their AI adoption strategies. The advantages include greater cost-efficiency, unified data access, and improved ease of use.

The next critical element is a unified data fabric. This creates a way to connect and govern all different data sources across hybrid and multi-cloud environments. Data fabric links and routes data from all connected sources to the tools and models that consume it. It also helps build unified data catalogs to discover available insights, implement and enforce necessary data privacy, security, and governance policies, plus ensure data replication across systems.

Speed matters significantly for multimodal AI applications. These systems need to make decisions within split seconds for customer experience optimization. A customer’s rating of a recent purchase or reporting of an issue must be processed and acted upon immediately. Real-time data processing pipelines, built with tools like Apache Kafka or Flink, enable quick capture of event-driven user signals and data synchronization across services.

Together, these building blocks enable data synchronization, governance, and interoperability at scale, which are necessary for effective multimodal AI deployment.

Cloud infrastructure considerations

Standalone natural language processing or computer vision models already require substantial cloud resources. Multimodal AI amplifies these requirements significantly. Many models require at least 16GB of VRAM for inference operations. You’ll need powerful multi-machine GPU clusters to train and run the models effectively.

Cloud service providers have already expanded their support for multimodal AI deployments. Google Cloud Platform offers access to NVIDIA B200 Tensor Core GPUs and Ironwood, plus managed services like Google Cloud Batch, Vertex AI Training, and GKE Autopilot that minimize the complexities of provisioning and orchestrating multi-GPU environments.

AWS provides access to 8 GPUs per instance, based on the latest NVIDIA architectures like H100, A100, V100, and Blackwell B200. The new AWS Neuron SDK helps optimize performance to deliver quantitative speedups and decreased latency.

Beyond raw computational power, multimodal AI requires seamless data synchronization. If image preprocessing or text embedding generation gets bottlenecked, the model won’t be able to produce quality outputs. This creates risks of modality incompleteness, when one or more modalities are missing during training or deployment, leading to skewed results.

 To avoid that, you’ll need to apply rigorous data pipeline optimization to keep data in sync:

• Enforce  consistent timestamps across all data streams to align events from text, audio, and video inputs 
• Embed metadata (e.g., session IDs, timestamps, source tags) directly into each data object to keep modalities contextually linked rather than siloed
• Use Apache Flink or Kafka Streams for real-time event correlation and time-windowed joins.
• Implement event versioning and schema enforcement to ensure compatibility between modalities
• Apply buffer windows or watermarking techniques to account for latency variations across modalities and restrict premature processing
• Configure observability tools to monitor ingestion lag and data skew for early issue detection 

Lastly, it helps to keep human‑in‑the‑loop validation, especially during the trending stage, to catch early failures and wire-in reliability before deployment.

Organizational readiness 

Deploying multimodal AI is not only about the technology, but also about the people. 

Talent access is a major constraint. The demand for AI experts fluent in multimodal data is outweighting the supply. Over 85% of companies struggle to hire AI developers, with average time-to-fill for open roles reaching 142 days in 2024. 

Even when you partner with experts in generative AI, internal organizational issues must be addressed. Beyond connecting systems, you’ll need to bridge collaboration gaps between teams, which are often the root cause of data silos. You’ll need to build a culture of proactive data sharing and clear data ownership, both of which require new habits of data democratization and data stewardship. Although challenging to implement, data stewardship can help both new and legacy businesses create stronger competitive advantages.

Unilever aims to operationalize over a hundred years of proprietary research — ancient formulations, trials, and consumer insights with AI models. As Alberto Prado, a global Head of Digital & Partnerships at Unilever, explained: “A century of skincare expertise is now structured, searchable, and ready to be applied. We are using AI to connect the dots across decades of research, accelerating discovery in new materials while simultaneously optimising formulations for specific needs, like different skin types”.  

The better news is that even newer entrants can achieve the same advantage by applying multimodal AI to previously untapped data — IoT sensor readings from oil and gas production monitoring systems, geospatial insights from drone mapping surveys, satellite imagery in precision agriculture, or behavioural data from self-service retail kiosks. Cordial, for instance, recently launched a multimodal AI model for optimizing retail brand messaging across brand creative, illustrations, photography, and text. Early customers reported a 2X to 3.2X increase in revenue. 

Effectively, multimodal AI can operationalize data sources, once considered too messy or unstructured to use at scale. But for that, you’ll need to align people, processes, and platforms around a common vision of data stewardship. 

Building a business case for multimodal AI adoption in marketing 

Multimodal AI will be a turning point for marketing personalization. It can help businesses recoup over $3 trillion, lost to data silos, by unlocking a new pane of contextual awareness, precision, and emotional intelligence, driven by a combo of first- and zero-party data. 

Getting there won’t be simple. Technical complexity, talent shortages, and cross-team collaboration gaps remain major roadblocks. Privacy and security add another layer of friction. Despite the hype, only 20% to 25% of organizations are actively using custom AI models in production today. 

That’s why a strong business case matters. You’ll need to connect the cost of data infrastructure optimization and model development against possible gains in conversion, retention, and operational efficiency. Moreover, it pays to think about wider, non-quantifiable benefits, like shorter feedback loops, faster campaign pivots, and a more profound understanding of your customers. And most importantly, risk management. By activating zero-party data, you reduce the compliance burden and reduce your reliance on third-party cookies. 

At Xenoss, we help organizations determine the optimal path for multimodal AI deployment. Our team can advise on infrastructure configuration, reducing cloud costs without compromising model performance. Our goal is to help you build both the business case and the technical system for AI that drives measurable ROI. Let’s discuss how multimodal AI can transform your marketing capabilities.

The post Multimodal AI in marketing: How zero-party data transforms customer personalization appeared first on Xenoss - AI and Data Software Development Company.

MAP abandonment rates are soaring as enterprises become more mindful of their tech spending in the era of AI and shifting customer demands. The Research In Action report reveals that 87% of companies are reassessing the functionality of their MAP, while only 14% are satisfied with their current one. The most common reasons for increasing MAP abandonment are the lack of innovation and the rising total cost of ownership (TCO). 

According to the Martech Intelligence Report, MAPs have been the most frequently replaced marketing software systems for five consecutive years. The share of companies that abandoned MAPs increased from 24% in 2023 to 40% in 2024, representing a nearly 70% rise. 

This blog post examines the key drivers behind the increasing MAP abandonment rates and unravels what’s next for enterprises considering MAP replacement. But does this mean that MAPs will soon cease to exist? Not exactly, at least when we take AI into the equation.

Three market forces that drive MAP abandonment among enterprises

Below are three major drivers that stimulate enterprises to abandon MAPs and seek other options to remain competitive.

Hyper-personalization with AI 

AI-infused MAPs have the potential to become a central hub for hyper-personalized marketing activities, reducing the need for costly integrations. Thus, the deciding factors for enterprises to abandon their current MAPs can be: 

• lack of full-fledged AI offerings
• simplified integration capabilities with AI tools
• a necessity for expensive workarounds to enable AI functionality

The findings from the State of Martech report prove that enterprises are already actively using AI in their daily operations:

• 54.2% of respondents use built-in AI assistants in their MAP and CEP as chatbots to communicate with customers and quickly find information.
• 72% of companies are already experimenting with AI for their marketing initiatives, applying it to many use cases (mostly content generation and ideation). 

But that’s only the tip of the iceberg; AI is capable of so much more. Jon Miller, a co-founder of Adobe Marketo, in his blog post on the future of B2B marketing, admits that starting from 2025, MAPs will be entirely reimagined with the help of AI. He claims that AI-native architectures are one of the most desired options. AI-native MAPs will enable users to ingest their enterprise data and then create and execute highly personalized marketing campaigns entirely without human assistance.

Composable Martech stacks

Composability is another trend fueling MAP abandonment, as it’s promising to reduce infrastructure costs and deliver quick and high ROI. This trend is gaining ground across the entire Martech, in particular, in the delivery of composable DXPs. In its essence, it’s the process of selecting best-in-class tech components and, through APIs, creating software architecture tailored to meet your unique marketing goals. Composability also inspired the emergence of low-code and no-code software, which reduces development time and costs while increasing flexibility. 

Therefore, the possibility of abandoning costly and monolithic MAPs to build a composable tech stack responsible only for essential MAP functions, with practically limitless integration possibilities, is going to change how enterprises perceive traditional MAPs. For instance, enterprises can implement smaller MAPs designed for SMBs but, with the help of additional integrations, achieve high performance at a lower cost than maintaining a single enterprise-grade MAP.

Vendor consolidation

Enterprises are reallocating their budgets and optimizing tech spending, which puts large-scale enterprise-level MAP systems under threat. Choosing only those vendors that bring the most value, or in other words, vendor consolidation, is one of the market fluctuations that encourages enterprises to switch to another MAP, develop it from scratch, or opt for tech composability. 

According to the State of Enterprise Tech Spending report, 76% of enterprises plan on pursuing vendor consolidation in 2025. Survey respondents also emphasized that before adopting any software solution, they’ll need vendors to provide precise specifications on ROI and business value. 

AI integrations, composable tech stacks, and vendor consolidation are mutually reinforcing market drivers. AI emphasizes optimization, cost reductions, and innovations. Technical composability and consolidation spring up to prepare the ground for the AI-driven marketing future. 

Let’s move from general market drivers to operational issues that encourage MAP abandonment.

Why enterprises abandon MAPs: The breaking points

Daily struggles with an outdated MAP tend to pile up until a marketing manager reaches their breaking point, and the frustration often sounds like this Adobe Marketo review: 

The product has served our business needs, but it’s been slow to keep up with features that its competitors roll out faster. I don’t see them as an innovator, theyʼre playing catch up.  

Where exactly do these systems fall short? Here are five reasons enterprises rip and replace their MAPs.

#1. Inability to actively measure ROI

Investments in the MAP software increase YoY, but enterprises don’t see the promised value. The need for real, tangible, measurable ROI is there. In 2011, the bar for MAP implementation success was relatively low; it was sufficient for the system to ingest historical data and build generic email templates. Now, implementing successful marketing automation software requires real-time data processing, big data analytics, AI integrations, and feature-rich interfaces. Therefore, the need for new MAPs arises, those that can provide innovation at a reasonable price.

The GNW Consulting survey reveals that for Martech implementations to be deemed successful, the time to ROI should be around one to six months. In contrast, for failing implementations, there is often no measurable ROI, or it takes more than a year to see real results. 

#2. Data centralization and integration issues

As specified by the Martech replacement survey, 47% of companies cite limited data centralization capabilities as one of the key factors driving marketing software replacement, in particular, MAPs. Data architectures that aren’t flexible enough to keep up with the increasing customer load disrupt marketing efforts. 

Modern enterprises need scalable data pipelines that collect data from multiple sources and enable insightful data analytics. With AI features appearing in MAPs, data issues are putting even higher pressure on enterprises, as it’s critical to ensure high data quality and security. The 2025 State of Your Stack survey reveals that data integration is the major Martech management challenge for 65.7% of respondents.

#3. Reporting and analytics constraints

If MAP vendors provide limited data analytics capabilities, enterprises miss out on valuable insights that might alter their marketing strategy. For instance, with proper data analytics in place, marketing specialists can validate the effectiveness of their ad campaigns and stop pouring money into those that prove ineffective. 

Traditional MAPs may have limited or insufficient data analytics functionality. For example, as suggested in this customer review on Adobe Marketo:

I’ve also found that reporting often needs a lift—it’s decent for quick overviews but underwhelming for deeper, customized insights. You’ll likely find yourself exporting data into Tableau or Python for serious analysis.

Deeper, customized insights are a key differentiator between innovative and traditional MAPs.

#4. Software integration issues

An outdated MAP system may hinder or complicate integration with modern tools, such as data enrichment tools and AI systems, limiting business growth. Lack of proper data synchronization between MAP and other essential marketing solutions (CRM, CDP, DXP, CMS) can also be an issue. 

Seamless integrations eliminate data silos, reduce errors, as the datasets are automatically ingested into the MAP, and allow for near real-time or real-time customer data exchange. This way, enterprises can build 360-degree customer overviews and achieve better alignment between marketing and sales teams to close deals effectively. 

#5. Failing cross-channel orchestration

Traditional email-focused MAPs are losing popularity, as there is a growing need for omnichannel functionality to track customers across all touchpoints, including social media, landing pages, SMS, webinars, physical stores, and chatbot conversations. Enterprises also point out that frequently missed features in MAPs include omnichannel and campaign personalization.

Consider this: a retailer installs AI-driven sensors in its stores to track customer behavior and analyze which product categories receive the most and least attention. This data can then be used to enrich marketing promotion campaigns for the least popular products and automatically distribute ads across multiple channels. With the help of AI, those messages can be targeted at specific customers based on their prior purchases, interests, and social media behavior. 

That’s what enterprises strive for when considering replacing their MAPs and ensuring a clear ROI: innovative marketing campaigns fueled by a modern tech stack and seamless data pipelines. Plus, they value the capability to measure the business impact of those campaigns with the help of advanced analytics tools. Let’s discuss the paths they choose to achieve such a level of marketing effectiveness.

What comes next: Post-abandonment strategies enterprises are adopting

What’s next after the decision to abandon your current MAP? The most common solutions could be migrating to another option that suits your goals much better, developing a custom MAP, or opting for composable stacks that substitute the MAP functionality.

Among MAP post-abandonment strategies, platform migration and custom development represent the most significant strategic decisions enterprises face. Each requires distinct considerations and investment levels.

Platform migration specifics

The most common post-abandonment strategy is switching to another MAP. But how long does it take to migrate from one platform to another? Jonathan Kent, Chief Revenue Officer at Hyperscale, puts it this way: 

Marketing automation platform migrations are complex, strategic initiatives that touch data, processes, integrations, compliance, and customer experience. Rushing through that in 30 (or even 60) days isn’t just unrealistic, it’s reckless. If a consulting firm promises to do it all in a month, that’s a red flag. If your internal team is in that situation, that’s a red flag too. The real solution? Plan ahead!

A general rule of thumb is to complete the platform migration within three to six months. This way, you can minimize disruption to your business workflow while ensuring a thorough migration of your key assets and data.

Here are key things to consider before, during, and after the migration period:

Step 1. Set clear goals and communicate them to stakeholders. Identify the key assets that need migration, outline how you will measure migration success, and assess all the risks associated with the migration process. Note that 58% of organizations that set measurable objectives for their Martech implementations and engage the C-suite throughout the process achieve ROI within six months.

Step 2. Compare different MAP vendors. To select a suitable MAP for migration, you should prioritize what’s most important for you (lead generation, data analytics, content optimization, AI/ML capabilities). You can create multiple comparison tables that list all the priorities for you, use industry reports that enumerate the pros and cons of different MAPs, or review customer reviews on G2 and Gartner.

Step 3. Audit data sources and break silos. For smooth data migration, locate all the necessary data across diverse sources and evaluate its quality dimensions (relevance, accuracy, integrity, timeliness, uniqueness). You should distinguish between the data that can be migrated (prospect records and activity, static lists) and the data that should be recreated (landing page templates, email templates).  

Step 4. Identify critical integrations. Explore the integrations your desired platform offers and identify which ones you’ll need to work on separately. You can then build automated data pipelines that enable quick data delivery between systems and tools.

Step 5. Iterative platform migration. Migrate your data and content assets in phases, testing and evaluating at every step. At this stage, it’s also critical to map data in the destination platform to ensure data formats align. For instance, prospect names in your current MAP may be listed in the format Full Name, whereas your destination MAP has a format of First Name, Last Name.

Step 6. Maintain parallel system functioning. During migration and for some time afterward, your MAPs should work simultaneously on different features to minimize disruptions to your currently running marketing campaigns.

MAP development from scratch

Developing your custom MAP is an option if none of the existing solutions on the market meet your needs. If that’s an issue, review key aspects of the custom development process to help you decide on the right track: 

High upfront investments. Custom solution development from scratch combines a best-of-breed tech stack to align completely with your business workflow, offering a potentially quicker ROI. But on the flipside, you’ll have to invest more time and money than you would in platform migration or optimization processes.

Skilled team composition. If you lack in-house tech specialists, you’ll need to assemble an expert team from external development professionals, preferably with Martech development expertise.

MAP feature map. Define the core features your MAP should include. The beauty of custom development is that the system feature set is entirely up to you. Include basic functionality such as lead management, nurturing, and campaign management, but also consider custom features like:

• deep-dive data analytics
• brand template designs for landing pages and emails
• native AI and ML capabilities trained on your enterprise datasets to make predictions and enrich marketing campaigns

MVP, PoC, and a full-fledged solution. Test the waters by deploying an MVP and PoC before proceeding to full-fledged development. In such a manner, you can iteratively add advanced features to the MAP and eliminate potential risks. 

Scalable and composable architecture. Designing a custom MAP architecture on AWS, Azure, or Google Cloud is the ultimate starting point to capture all the relevant data (including structured and unstructured) from multiple marketing sources. 

At the architecture stage, you can set up real-time and batch big data processing to handle increasing loads and funnel it into custom data analytics tools. To enable automated decision-making at scale, which is crucial for MAP systems, event-driven architecture serves as a foundation for real-time data pipelines. 

By combining patterns of event-driven architectures (e.g., message queues such as Kafka, RabbitMQ, AWS EventBridge) with composable microservices architecture elements (e.g., orchestration tools such as Camunda, Temporal, and Apache Airflow), you can achieve enhanced system performance, resilience, scalability, and stability during periods of peak demand.

Privacy- and security-by-design. Prioritize security and compliance with GDPR, CCPA, PCI DSS, HIPAA, or any other industry-specific regulations, laws, and standards from day one. A skilled development team considers role-based access controls, data encryption, anonymization, and pseudonymization strategies during the planning and design stage.

Success metrics: What works in post-abandonment implementations

What’s most important after you start working with your new MAP is consistently tracking 

success. For that purpose, focus on:

• Custom KPIs. Depending on the goals you set before MAP migration or development, you should define your unique KPIs to measure (such as conversion rate and open or click rates) to evaluate your success. Additionally, track over time whether your marketing team is using all the platform features and whether your feature set is sufficient or insufficient.
• (Responsible, accountable, consulted, informed) RACI matrix. By developing a comprehensive RACI matrix, which identifies who is responsible for different processes within your migration or development project, you can ensure accountability when measuring the project’s success.
• Team training. The top lessons from failed Martech rollouts include the need for comprehensive team training (66% of respondents). Help your team grasp the essence of the new platform more quickly and deliver the expected results.
• Change management strategies. Organizations reporting fewer challenges with change management and cross-functional misalignment are more likely to achieve faster and longer-lasting adoption of their new MAP. Establish a deep-rooted culture of change within your organization by clearly communicating it as a strategic vision and supporting your team throughout the change process.

Building or migrating is a question of your current priorities. Define the core reason for MAP abandonment first. Is it a data integration issue? You can resolve it by auditing your enterprise data management processes and migrating to a more mature MAP platform that employs advanced data orchestration techniques.  

Or is it a lack of advanced functionality, integrations, and AI capabilities that’s triggering your MAP abandonment? Then your go-to option can be custom MAP development or a composable tech stack. Discover from the case studies in the next section how different organizations approached MAP abandonment.

Enterprise abandonment case studies: When MAPs fail beyond repair

Among 200 people surveyed by GNW Consulting, 31% of respondents admitted that their recent Martech implementations had failed or delivered neutral results. The major risks of failure are:

• Revenue loss. A higher average contract value (ACV) combined with the purchase of costly technology equals a greater risk for a business. Thus, to achieve post-implementation ROI, enterprises should carefully plan their MAP migration, development, or optimization strategies.
• Team churn. Failed MAP implementations can result in marketing specialists either quitting or being fired; 42% of respondents reported that failed implementations led to employee turnover. 

To avoid such outcomes, the following companies have thoroughly evaluated their as-is state to define which MAP migration strategy would be most appropriate for their businesses.

Case #1. MAP migration combined with vendor consolidation 

A US-based SaaS company, offering design and brand management services, decided to migrate from Salesforce and Adobe Marketo to HubSpot. The company needed eight FTEs to administer their growing tech stack and still struggled with cross-functional misalignment between sales, marketing, and operations teams. Here are the specifics of this migration project: 

• Timeline: The project took 90 days to implement a new MAP and consolidate all Martech tools.
• Strategy: The company adopted HubSpot’s Marketing, Sales, Service, and Operations Hubs to replace its legacy stack.
• Lessons learned: 
  • The technology stack should align with the business workflow, not vice versa
  • Tool sprawl increases costs and decreases business value
• Results of MAP migration and vendor consolidation: 
  • Around $77,000 saved annually
  • 50% reduction in technology expenses thanks to the consolidation of sales, operations, and marketing solutions
  • Faster sales cycles

Case #2. From traditional MAP to composability

Another US-based B2B SaaS organization that provides diverse AI development services, abandoned Adobe Marketo and built a composable stack instead. Their marketing team had issues with the product’s usability, slow processing speed, and obsolete data management approaches. Let’s take a closer look:

• Timeline: Abandoned costly Marketo license in 2023 and launched a composable setup in less than 2 weeks.
• Strategy: Built a modular stack, consisting of: Iterable (messaging), Contentful (assets), Tray.io (workflow), Salesforce CRM.
• Lessons learned: 
  • Monolithic MAPs have limited flexibility
  • Composable systems require stronger RevOps but deliver enhanced agility
• Results of building a composable stack: 
  • Campaign personalization improved
  • Messaging costs were 10 times lower 
  • Marketers gained speed and control

These cases demonstrate that the era of Frankensteined Martech stacks and monolithic MAPs is coming to an end, with modern companies prioritizing omnichannel personalization, usability, and data processing speed. Enterprises that evaluate their current workflows, clean up data debt, and choose scalable, integration-ready tools are better positioned to avoid revenue loss and retain top talent. 

Navigating the MAP market’s future

In the near future, the MAP market will be completely redefined. Traditional MAPs will either have to adapt and change fast or experience increasing customer churn. Composability, AI-enabled hyper-personalization, and vendor consolidation are three key pillars that drive market progress. We’ll likely witness the emergence of new MAP vendors with AI-native architectures, seamless cross-channel orchestration and personalization, and enhanced capabilities for data and third-party integrations. And they won’t cost an arm and a leg while offering so much more than legacy MAPs from the 90s. 

To remain competitive and make marketing a strategic driver at your organization, start taking the first steps towards modernizing and consolidating your Martech stack, with MAP at its center. Analyze how your MAP contributes to marketing outcomes, highlight inefficiencies, and outline a transformation roadmap that shifts your stack from rigid and monolithic to modular and insight-driven.

The post Marketing automation platform breakdown: Why enterprises abandon their MAP implementations and what comes next appeared first on Xenoss - AI and Data Software Development Company.

Because most enterprise MarTech stacks are highly fragmented. Sixty-six percent of teams are using sixteen or more different MarTech tools, and most of them don’t talk to each other. Content lives in one silo, user data in another, and personalization logic…well, that’s often hardcoded by a team that left two years ago.

This fragmentation destroys ROI. With technology utilization rates at just 33%, two-thirds of enterprise MarTech investments become budget waste. Organizations lose millions on overlapping software licenses, duplicated workflows, and tools that teams never fully adopt.

The solution lies in a modular MarTech architecture that combines composable DXPs with real-time CDPs. This approach finally delivers the seamless, personalized experiences that marketers have promised and customers expect.

The new stack: How composable DXP and CDP enable better enterprise personalization 

Years of scattered MarTech investments have left enterprises with a brittle ecosystem — one where customer profiles are outdated, content is trapped in channel-specific workflows, and launching new experiences means weeks of integration work. Only 17% of leaders say their marketing tech stack works well together.

That’s why many are turning to modular stack rationalization. It’s not just a cost play. Simplifying your stack improves governance, reduces complexity, and unlocks new personalization capabilities, especially when paired with real-time customer data and API-first content delivery.  

Adobe’s research shows impressive ROI gains when companies consolidate their scattered point solutions and get their customer data working together.

To enable real-time customer experience personalization, companies should look into a composable MarTech architecture where:

• A headless, API-first DXP renders content dynamically
• A real-time CDP ingests behavioral signals and syncs customer state

And everything flows through lightweight data pipelines designed for real-time streaming and fast activation. The result is personalization that scales with your business, delivers measurable results, and adapts as your needs evolve.

“By 2026, at least 70% of organizations will be mandated to acquire composable DXP technology, as opposed to monolithic DXP suites, compared to 50% in 2023. Gartner ”

Under the hood of a composable experience stack 

Traditional CDPs were built to solve one problem: collecting customer data in a central location. Some platforms added data enrichment features later, but real-time experience personalization? That was never part of the original design.

Legacy DXPs have their own issues. They’re stuck with channel-specific workflows and outdated CMS architectures that create bottlenecks instead of enabling smooth personalization.

Composable CDPs and DXPs take a different approach. They’re designed from the ground up for real-time data flows and immediate action on customer signals.

What composable CDPs handle

Composable CDPs manage the ingestion, processing, and activation of customer data at speed and scale:

• Behavioral signal capture: Clicks, views, purchases, and engagement events
• Identity resolution: Creating unified customer profiles across devices and touchpoints
• Real-time segmentation: Dynamic audience updates that sync instantly with other systems

Key technical components:

• Real-time data pipelines (Kafka, Flink, Pulsar)
• Identity resolution engines with first-party ID graphs and cross-device matching
• Customer profiling and dynamic segmentation tools
• Behavioral event enrichment for clickstream, transaction, and pageview data
• Privacy compliance layers for GDPR/CCPA consent management
• Predictive scoring models for churn risk, lifetime value, and purchase intent
• Event stream versioning with automated contract testing

What composable DXPs handle

Composable DXPs are responsible for delivering dynamic content experiences across channels:

• API-driven content activation: Dynamic content delivery based on real-time context
• Millisecond experience rendering: Personalized pages and components that load instantly
• Cross-channel delivery: Consistent experiences across web, mobile, email, and in-app environments

Key technical components:

• Headless CMS with API-first content management
• Experience orchestration engine handling rules, triggers, and personalization logic
• Visual content editor for non-technical team members
• Personalization API layer that integrates directly with CDP customer data
• Channel-agnostic content delivery APIs
• Content versioning and localization management
• Built-in experimentation framework for A/B testing and multivariate optimization
• Edge delivery network with intelligent caching and real-time cache invalidation

Integration layer 

The integration layer connects these systems through a message bus (Kafka, Google Pub/Sub), standardized APIs, and shared data contracts that ensure reliable communication between components.

To better understand how composable DXPs and CDPs slot together, let’s look at each system layer. 

Real-time data layer: Pipelines powering dynamic personalization

In any composable experience stack, the real-time data layer is what turns static content into dynamic personalization. Think of it as the nervous system connecting what a user just did to what your interface does next. Without real-time data flows, your DXP operates blindly while your CDP reacts hours too late.

Well-designed real-time data pipelines let you respond while users are still actively engaged, rather than in tomorrow’s batch job. This enables powerful use cases: cart abandonment triggers that update homepage banners before visitors leave, geo-targeted offers that adapt as customers move between locations, or instant product recommendations after someone views an item.

To support this kind of immediacy, your pipeline architecture must be fast, fault-tolerant, and schema-aware. Xenoss data engineering team recommends a modern stack built on:

• Kafka for blazing-fast event streaming
• Flink for real-time stream processing and transformations
• Debezium for change data capture (CDC) from transactional databases
• CDC-enabled DBs like PostgreSQL or MySQL to keep systems in sync with minimal lag

Additionally, you’ll need to carefully consider how stream management is handled. Schema versioning ensures updates don’t break downstream services. Retry logic helps manage processing failures gracefully. Late event handling ensures that delayed data is still processed correctly.

Real-time data management infrastructure is a cornerstone of advanced personalization. Without the ability to capture, process, and act on events instantly, your MarTech stack might still be modular, but it won’t be intelligent.

Pro tip: Real-time data and composable architectures enable more AI use cases 

Composable MarTech stacks create the ideal foundation for deploying AI models. With modular APIs and event-driven pipelines, real-time data flows seamlessly from customer touchpoints into both CDPs and DXPs, enabling sophisticated AI applications that were impossible with legacy architectures.

This infrastructure supports a wide range of AI-powered personalization capabilities:

• Behavioral micro-segmentation based on real-time interaction patterns
• Context-aware content ranking that adapts to individual preferences
• Lookalike audience generation for campaign expansion
• Dynamic creative optimization that personalizes ad content in real-time
• Win-back campaign automation triggered by engagement signals
• Adaptive journey orchestration that adjusts based on customer behavior
• Intent-based product recommendations driven by browsing patterns
• Predictive lead scoring using multi-channel engagement data
• Real-time sentiment analysis across customer communications
• Advanced anomaly detection for fraud prevention and quality assurance

The key advantage of composable architectures lies in their modularity. Different AI models can be developed, tested, and deployed independently without disrupting existing systems. This approach accelerates experimentation while reducing integration complexity and deployment risk.

Real-world impact: Xenoss helped a leading CEE retail marketplace launch ML models that automatically optimize RTB campaigns based on user behavior signals. Each model is pre-trained for a specific shopper segment and triggered when a merchant launches a new campaign. This real-time intelligence led to 40–50% lower CPCs and a 24% drop in customer acquisition costs.

How to synchronize content and customer data across systems

Your DXP has to be in sync with other elements in your stack: CDP, CMS, CRM, and MAP, to serve experiences based on up-to-date context. 

The foundation of reliable integration starts with data schema harmonization. Without unified data models, personalization logic breaks down, and customer states become inconsistent across channels.

The first step involves establishing a unified customer identity. Create consistent customer IDs using hashed email addresses, loyalty program identifiers, or device fingerprints. Normalize data formats, naming conventions, and taxonomies across all connected systems. Implement a central schema registry to manage data models, consent flags, and privacy metadata consistently across your entire stack.

API reliability requires equal attention. Design APIs with built-in retry logic, circuit-breaker patterns, and specific latency targets (typically 50ms or less for personalization use cases). Build fallback mechanisms that keep customer experiences functional during partial system outages. Use contract testing and automated event versioning to maintain integration stability through system updates and feature releases.

Real-time behavioral data creates the intelligence layer. A well-designed feature store captures fresh customer signals such as recent category views, discount sensitivity scores, and cart abandonment risk indicators, enabling your DXP to deliver personalized content within milliseconds without hardcoding business logic into frontend applications.

Implementation example: GetYourGuide used Contentstack’s composable DXP with a headless CMS, APIs, and contentors to enable data-driven content delivery on their website. The new setup allowed them to deliver personalized content to their 500,000 million user base 90% faster, regardless of the device. 

How composable DXPs deliver personalized experiences across channels

Composable DXPs eliminate cross-channel content silos. Instead of manually replicating experiences for each touchpoint, you can leverage centralized orchestration and decentralized delivery features to launch adaptive content everywhere at once.

The experience API gateway: Your personalization hub

The experience API gateway serves as the central intelligence layer connecting backend systems with frontend channels. When a customer visits your website or opens your mobile app, the frontend captures their interaction and sends a request to the API gateway along with contextual signals like location, device type, user ID, and session information.

The gateway parses those tokens and routes the request to the relevant backend services:

• The Content Management System (CMS) returns modular content blocks
• The Customer Data Platform (CDP) provides user profiles and segmentation data
• The decision engine determines which personalization logic or AI model to apply

Using these inputs, the API gateway assembles a personalized experience, including product carousels tailored to browsing history, localized promotional banners, and AI-generated headlines that match individual preferences.

Stateless rendering for instant adaptation

Stateless rendering engines handle content delivery by making dynamic calls to the API gateway using current context tokens. These engines don’t store session data locally. Instead, they retrieve fresh inputs like current location, device capabilities, or recent customer actions to request the most relevant content from backend systems.

This stateless approach delivers two key benefits: improved system scalability and instant experience adaptation based on customer behavior, all without requiring page reloads or app refreshes.

Cross-device experience continuity

Context synchronization ensures seamless experiences as customers move between devices. Composable DXPs combine first-party data (email logins, loyalty program IDs, purchase history) with third-party signals (cookies, advertising network IDs, device fingerprints) to maintain consistent customer context. When someone switches from mobile browsing to desktop checkout, their personalized experience continues exactly where they left off.

Intelligent caching with real-time invalidation maintains fast performance while ensuring content accuracy. Frequently requested assets are cached at edge locations or client devices to reduce latency. When context changes, such as updated pricing, inventory levels, or promotional campaigns, cache invalidation happens instantly to prevent outdated content delivery.

This architecture enables dynamic pricing updates, real-time inventory displays, and geo-specific content without sacrificing performance or user experience quality.

Building trust through data governance and security monitoring

API-driven personalization introduces new requirements for data security, privacy compliance, and customer trust. While customers value personalized experiences, they expect transparency about how their data gets collected and used.

Current trust levels reveal significant gaps. Only 30% of US consumers trust brands with their personal data. However, 66% would reconsider their stance if companies provided clearer transparency about data collection practices and usage purposes. Regulatory pressure continues to intensify as governments worldwide implement stricter privacy controls and compliance requirements.

Essential security practices for composable MarTech

Modular architectures require comprehensive security measures that protect data across multiple system boundaries. The foundation starts with access control and authentication, implementing role-based permissions with granular control over each service. Token-based authentication secures all API interactions between systems, following the principle of least privilege across system integrations.

Comprehensive audit logging forms the accountability layer. Every data access event and content delivery decision gets logged across connected services, ensuring personalization respects embedded consent flags and privacy preferences. These detailed audit trails support both compliance reporting and security investigations when issues arise.

Data classification and protection require schema-level sensitivity tags that govern data usage across connected systems. Personally identifiable information gets automatically redacted before exposure to analytics tools, while data masking protects non-production environments and testing workflows from exposure to sensitive customer data.

Real-time monitoring and alerting provide an early warning system for potential issues. Automated monitoring tracks integration failures and CDP-DXP data conflicts, while alerts flag unusual access patterns or potential security incidents. API performance and error rate monitoring across all system connections helps maintain both security and performance standards.

Coordinated data lifecycle management ensures compliance without operational chaos. Each service maintains its own retention policies, while central orchestration layers or data catalogs prevent conflicts and data loss. Automated deletion processes handle privacy regulation requirements while vendor vetting evaluates every new technology component for encryption standards, data residency requirements, and consent management capabilities.

Composable architectures thrive on speed and flexibility, but neither should come at a cost of data negligence. With the right security monitoring, data governance, and fallback mechanisms in place, you can maintain scalable systems without jeopardizing customer trust. 

Business results: What composable DXPs unlock for enterprises

Composable DXPs enable enterprises to respond to market demands without constant replatforming cycles. Real-time data pipelines and modular APIs accelerate personalization deployment across channels while eliminating complex hardcoding requirements that slow traditional implementations.

Organizations adopting composable experience architectures achieve faster experimentation cycles, improved conversion rates, and stronger ROI from marketing campaigns. As AI agents become standard in marketing technology stacks, composable DXPs provide the clean integration points needed to deploy next-generation models without rebuilding core infrastructure.

Companies implementing these architectures typically see faster time-to-market for new experiences, improved cross-channel consistency, and enhanced personalization capabilities that drive higher engagement rates. The modularity advantage becomes particularly valuable during technology evolution—enterprises can integrate new capabilities without disrupting existing workflows or replacing entire platforms.

Xenoss data engineering teams specialize in building the infrastructure that powers composable marketing transformation. Our expertise spans real-time data pipeline design, API orchestration between CDP and DXP layers, and custom event-driven personalization systems.

Ready to transform your marketing technology infrastructure? Contact our team to get a free composable DXP architecture assessment and roadmap tailored to your current MarTech stack and personalization goals.

The post Composable DXPs: How to unify content and customer data for real-time enterprise personalization appeared first on Xenoss - AI and Data Software Development Company.

In our previous deep dive, “Disrupting the MarTech stack: How generative AI is reshaping traditional marketing,”  we explored the breakdown of once-reliable marketing channels. SEO, social campaigns, email blasts, paid media, and influencer outreach are all buckling under the combined pressure of saturation, rising costs, and platform hostility. As 2025 unfolds, it’s clear: the playbooks that once delivered reach have turned into background noise. Users, overwhelmed by irrelevant messages, scroll past without pause.

Take SEO — once a cornerstone, now a rigged game. Brands spend arduous months optimizing content, only to find themselves outranked by informal Reddit threads or summarily penalized by Google’s relentless algorithm tweaks. 

For instance, News Corp’s flagship UK brand, The Sun, witnessed a devastating 50% loss of organic traffic in late 2024, a direct consequence of Google’s algorithm changes that explicitly punished broad, listicle-style publishing in favor of community discussions. 

The New York Post experienced a 27% traffic decline during the same period, signaling not isolated blips but a structural reinterpretation of authority by Google, shifting away from high-volume generalist publishers. 

Even HubSpot, once the quintessential blueprint for B2B content marketing, suffered a dramatic loss in visibility across thousands of high-intent queries, with its vast archive of SEO-first content, often thin on topical authority, deprioritized by the Helpful Content Update. Hubspot’s internal metrics suggest up to a 40% decline in organic leads from Google since December 2024.

Independent publishers have suffered even more. DMARGE, an Australian lifestyle site, spent $200,000 to regain lost traffic, only to see monthly visits crash from 8 million to 300,000. Programmatic revenue shrank to 3–5% of previous levels. With editorial cuts and falling ROI, the team shut down its site and shifted to Instagram and newsletters. Google no longer rewards independent content; it buries it.

And it’s not just SEO. Most scalable channels are in decline:

• Influencer marketing burns budgets for minimal return.
• Email campaigns struggle with low deliverability and engagement.
• Paid media faces rising CPMs and rampant click fraud.
• Affiliate and referral programs are bogged down by fraud and admin overhead.
• PR might build awareness, but rarely converts.

The aggregated effect is a landscape where everything that once worked has been diluted into mere noise, leaving users disengaged and exasperated.

AI discovery: Why conversations are replacing search

As traditional marketing channels deteriorate, a new discovery layer is rapidly taking shape: generative AI. Tools like ChatGPT, Perplexity, and Microsoft Copilot are transforming the way users interact with information. Discovery no longer begins with fragmented keywords in a search bar. It begins with context-aware, interactive conversations. A query like “best credit cards for travel” becomes an ongoing dialogue shaped by prior inputs and refined recommendations.

This shift is producing measurable changes in user behavior. According to Adobe Analytics, traffic from AI referrals consistently outperforms traditional sources across key engagement metrics. Users spend 8% more time on site, view 12% more pages, and have a 23% lower bounce rate. These figures point to higher-quality, more engaged sessions.

Engagement growth is matched by traffic acceleration. During the 2024 holiday season, Adobe recorded a 1,300% year-over-year increase in AI-driven retail visits, with Cyber Monday traffic rising by 1,950%. By February 2025, AI referral traffic was up 1,200% compared to just seven months earlier. Similar trends are playing out across other sectors, with banking and travel seeing 1,200% and 1,700% increases, respectively. While AI referrals still represent a smaller share of total traffic than paid search or email, their rapid expansion signals a major realignment in digital discovery.

The driver behind this trend is intent. Semrush’s comparison of ChatGPT and Google reveals a distinct behavioral difference. Over 52% of ChatGPT queries are informational, compared to 36.4% on Google. Navigational queries dominate Google at nearly 50%, but drop to just 34% on ChatGPT. Users interacting with AI systems are seeking answers, evaluating options, and engaging in multi-step problem solving. Generative AI is not simply modifying how people search. It is redefining the entire process of exploration and decision-making online.

AI-native ads: How monetization works inside conversational interfaces

As this shift accelerates, platforms adapt to a new user engagement model. ChatGPT now routes traffic through homepage-weighted links, while Microsoft Copilot creates personalized session flows. Analytics providers like Similarweb are building the infrastructure to track and interpret AI-originated traffic. They signal a fundamental reordering of how users discover, evaluate, and act online.

Discovery no longer flows through SEO, email, or influencer traffic as it once did. Generative AI introduces a new interface with its own logic and mechanics. Brands that understand how these systems interpret, rank, and surface information will gain a lasting competitive edge.

But visibility is only step one. The larger opportunity is monetization, embedded directly inside the interface. AI-native ad formats are beginning to surface within conversations, APIs, and assistant workflows. The interface itself becomes the inventory.

In this model, attention isn’t measured by impressions. It’s measured by engagement: follow-up questions, real-time filters, co-piloted decision flows, and conversational prompts. Display banners are being replaced by dialogue.

Here’s how this new advertising layer is taking shape in 2025.

Microsoft’s Copilot: Ad voice, compare & decide ads, and showroom experiences

Microsoft is moving fast to integrate advertising into Copilot’s conversational environment. Its “Ad Voice” feature is designed to embed sponsored content naturally into the flow of conversation, guiding users from query to recommendation without breaking context. The ads are labeled and disclosed, but presented as a native part of the user’s interaction.

Another emerging format is the “Showroom ad,” which transforms the assistant into a virtual product advisor. A question like “What’s a good running shoe for flat feet?” triggers a dynamic filtering experience, surfacing tailored options based on user-specified criteria such as “eco-friendly,” “under $100,” or “waterproof.”

Another emerging format is the “Showroom ad,” which transforms the assistant into a virtual product advisor. A question like “What’s a good running shoe for flat feet?” triggers a dynamic filtering experience, surfacing tailored options based on user-specified criteria such as “eco-friendly,” “under $100,” or “waterproof.”

Early results are strong. Copilot search ads outperform traditional search placements, with a 25% improvement in ad relevance metrics and a 76% higher conversion rate. These formats are already live within Performance Max (PMax) campaigns, meaning advertisers using PMax are automatically reaching Copilot users. Adoption is particularly relevant for Gen Z, who now represent over 30% of the Copilot mobile user base, and are reshaping digital commerce behaviors in real time.

Amazon’s Rufus: Contextual product recommendations

As of January 2025, Amazon has begun embedding Sponsored Ads directly into Rufus, its generative AI shopping assistant. Designed to deliver highly personalized product recommendations, Rufus analyzes product metadata, customer Q&As, reviews, and images to understand intent and serve contextually relevant results.

Rufus now appears above the fold on desktop and ranks higher on product pages, increasing the impact of AI-aware listing strategies. The assistant functions like a product analyst, adapting to user behavior such as recent searches, click-throughs, purchase history, and demographic trends.

For sellers, success requires listings that are structured for AI consumption. That includes using diverse keyword types: intent-driven, solution-specific, quality-descriptive, and comparison-based. Relevance is determined through context signals, not just keyword density.

Snapchat’s MyAI: Personalized engagement and adaptive ad formats

Public details on Snapchat’s MyAI ad placements remain limited, but the assistant is a key part of the company’s broader “AI-powered advertiser solutions,” introduced in May 2025. The focus is on improving content ranking and personalization using advanced AI and machine learning models.

Recent additions include “Smart Campaign Solutions,” which feature Smart Bidding and Smart Budget tools to optimize campaign performance. Snapchat has also rolled out new ad formats such as “First Snap,” a single-day takeover that appears in Chat feeds and opens as a full-screen video, and “Web and App Auction Ads,” which expand inventory across platforms.

MyAI’s personalization capabilities suggest a future role in mid-dialogue product suggestions and integrated sponsored content. As Snapchat invests in longer-form content and creator monetization, the opportunity grows for brands to embed messaging directly into user-generated interactions without disrupting the experience.

Perplexity AI: Comet browser and behavioral data targeting

Perplexity AI’s monetization strategy has become a focal point for privacy concerns in 2025. The company plans to collect extensive behavioral data through its upcoming Comet browser to build highly personalized advertising profiles. While this model supports more precise ad targeting, it has drawn scrutiny for its scope and potential privacy trade-offs.

CEO Aravind Srinivas has confirmed that Perplexity aims to capture data beyond direct AI queries. The objective is to construct rich user profiles based on browsing patterns, content interactions, and inferred intent. This data will likely fuel ad placements in Perplexity’s “Discover feed.”

The assistant’s conversational flow also plays a role. Follow-up prompts within chats allow the system to map intent progression across queries, creating additional context signals for ad delivery. This behavioral layer gives Perplexity an edge in targeting, but also intensifies the debate around surveillance-level data collection.

Perplexity’s Comet browser, set to launch in 2025, is designed to collect user data far beyond chat-based interactions. CEO Aravind Srinivas has stated that the goal is to capture comprehensive behavioral patterns, including purchase history, browsing habits, and location-specific preferences, to build richer advertising profiles.

While some AI use cases remain task-focused, Srinivas emphasized that deeper personalization comes from everyday behavioral signals. Comet’s infrastructure is explicitly intended to gather this data across contexts, reinforcing the company’s strategy to power hyper-targeted advertising through full-spectrum user profiling.

The broader insight: Advertising is native, contextual, and deeply personal

The overarching insight that advertisements are moving intrinsically within the product experience is affirmed and rapidly accelerating. This profound evolution is driven by AI’s unparalleled ability to comprehend deep context and intent, facilitating highly personalized and often subtly integrated ad placements.

Hyper-personalization and conversational commerce

AI enables dynamic segmentation based on real-time engagement patterns, creating truly personalized experiences across every touchpoint. This capability transcends static demographics, delving into subtle behavioral insights. Chatbots are explicitly transforming into tools for “conversational advertising,” adeptly capturing detailed user queries to yield insights into consumer intent that far surpass traditional search algorithms.

Seamless integration and native formats

The prevailing focus is on embedding ads directly into content environments. Emerging formats like in-scene media and virtual product placement enable brands to integrate seamlessly into videos, shows, and interactive experiences. For example, Warner Bros. Discovery’s “Shop with Max” technology allows streamers to purchase products marketed on select programming in real time through a curated second-screen experience. 

This uses QR codes, audio, and visual cues to identify relevant themes and on-screen elements, allowing marketers to reach audiences engaged with specific content topics. The ultimate objective is for advertisements to feel like an organic part of the viewing experience, rather than an intrusive disruption.

Ethical and privacy considerations: The dawn of surveillance 2.0

The rise of generative AI has amplified long-standing concerns around data privacy. General-purpose AI models carry inherent risks, including memorization of training data and limited user control over personal information. Challenges persist in fully deleting user data, even under mandates like GDPR and CCPA. The opacity of model behavior remains a critical issue, often referred to as the “black box” problem. A growing concern is the repurposing of personal data, using it in ways beyond the original scope, which raises questions about consent, legality, and user trust.

As AI assistants take on more contextual and persistent roles, the depth of data they collect surpasses anything enabled by cookies. Emotional tone, behavioral signals, and even implied intent are captured through ongoing interactions. This creates a powerful targeting framework and a potential surveillance infrastructure. The industry is entering what many now call “Surveillance 2.0,” with increasing pressure to implement privacy-by-design principles and offer users more visibility and control.

Technological safeguards are advancing, including PII redaction, synthetic data, differential privacy, confidential computing, and cryptographic protections. However, most involve a trade-off between privacy and utility. Reduced access to granular data can impact model accuracy, particularly in general-purpose systems. Striking a balance between personalization and privacy will remain a central challenge in AI deployment and regulation moving forward.

AI agents with memory and emotional intelligence

Modern AI assistants are now designed to retain context across sessions, enabling follow-ups, reminders, and long-term personalization. Many are also being trained to detect tone, mood, and even stress levels, allowing for more empathetic and adaptive interactions. This level of contextual awareness supports the delivery of highly relevant mid-conversation ads that align with user intent without breaking flow.

In 2025, AI-driven embedded advertising has entered a phase of rapid maturity. Conversation itself has become the interface and the ad unit. This creates powerful new opportunities for brands to engage users through personalized, context-aware experiences. At the same time, it raises critical questions about privacy, data ownership, and responsible system design. The tension between innovation and regulation is now a defining factor in the future of AI-powered commerce.

Navigating the new advertising architecture

Generative AI transforms advertising from an external layer into a core system function. Interfaces have become intelligent environments where discovery, evaluation, and monetization occur simultaneously, often without users realizing they’ve left the search phase. This new architecture is driven by systems that analyze context in real time and act on behavioral signals with increasing precision.

As conversational interfaces become the primary gateway to information, the economics of attention are shifting. Visibility now depends on alignment with AI reasoning: how systems interpret relevance, confidence, and commercial intent. Brands and platforms must learn to operate within these logic models or risk losing their position in the discovery flow.

The regulatory landscape will lag behind technical capability, but pressure is mounting. Consent frameworks, auditability, and data minimization will become critical components of any scalable strategy. The companies that treat trust, explainability, and user control as product features will be best positioned to lead.

What’s emerging is not just a new channel, but a new foundation for how influence is built and value is exchanged. Advertising is no longer a message delivered within content. It is a responsive, context-aware layer inside the interaction itself, and its rules are still being written.

The post The new advertising layer: How AI turns conversations into commerce and rewrites privacy rules appeared first on Xenoss - AI and Data Software Development Company.

And just as that decay becomes impossible to ignore, a new channel is quietly gaining ground: generative AI. Tools like ChatGPT, Perplexity, and Copilot aren’t just changing how people search; they’re starting to replace search altogether.

Why traditional marketing channels are no longer sustainable

There was a time when organic reach and a bit of cleverness could take you far. That time is over.

SEO: The long game 

Let’s start with SEO. Once the cornerstone of long-term marketing strategy, it now feels like a rigged game. You spend months optimizing content, only to be outranked by Reddit threads or slapped down by Google’s algorithm tweaks. And if you’re publishing anything even remotely outside your “core expertise,” you’re toast.

To make matters worse, SEO is a long game. According to a poll by Ahrefs with 3,680 responses on LinkedIn and X, it typically takes three to six months to see results. So not only is it increasingly unstable, it’s also painfully slow, leaving brands vulnerable to algorithm swings in the meantime.

Consider News Corp: In late 2024, their flagship UK brand, The Sun, lost 50% of its organic traffic, as revealed in their Q4 earnings report. Analysts linked the drop to multiple Google algorithm changes that punished broad, listicle-style publishing and gave preference to Reddit discussions. Another News Corp publication, the New York Post, saw a 27% traffic decline during the same period. These were not isolated blips; they reflected a structural change in how Google was interpreting authority, with a shift away from high-volume generalist publishers.

HubSpot didn’t fare any better. Often seen as the blueprint for B2B content marketing, the company suffered a dramatic decline in visibility across thousands of high-intent queries. According to search consultants, HubSpot’s downfall was tied to its vast archive of SEO-first content, material optimized to attract traffic but thin on topical authority. The Helpful Content Update appeared to deprioritize such strategies, favoring brands with deeper, narrower expertise instead. HubSpot’s own traffic metrics suggest a decline of up to 40% in organic leads from Google since December 2024.

Independent publishers got it worse. DMARGE, an Australian men’s lifestyle publication operating since 2009, spent $200,000 trying to recover from algorithm changes, experiencing a drop from 8 million to 300,000 monthly visitors. The collapse was tied to Google’s November 2021 Core Update and intensified with the rollout of the Helpful Content Update. Despite employing a 10-person editorial team covering topics like watches, cars, food, and travel, the site saw no manual actions or direct warnings from Google.

The cost of recovery was staggering:

• $25,000 for SEO consultants
• $120,000 for web development
• $75,000 for content creation and editing

Factoring in the owner’s own time, calculated at $100/hour over three years, the total investment approached $295,000. Technical changes included extensive testing of permalink structures, server configurations, and Core Web Vitals optimization, with $60,000 to $70,000 invested annually in development alone.

Programmatic ad revenue fell to just 3–5% of prior levels, forcing editorial cuts and compromising content quality. The publisher ultimately concluded that maintaining a traditional website became commercially unviable. They now see better content reach through Instagram and newsletters than via their own domain.

The DMARGE case laid bare a deeper truth: Google’s ecosystem no longer rewards independent publishers. It punishes them.

You can blame the Helpful Content Update, but the trend started earlier. The law of diminishing returns is now the law of the land. Every scalable channel has been overfished:

• Influencer marketing: Big spend, tiny ROI. Once the traffic spike fades, nothing sticks.
• Email: Ends up in spam, low open rates, and even lower CTRs.
• Paid media: Brutal CPM inflation, rampant fraud, investor backlash.
• Affiliate/referral: Ripe with fraud, heavy admin load, and surprisingly poor results.
• PR: Good for buzz, terrible for performance. Your competitors get the same mention next month.

Everything that used to work is now just noise. Users are burned out. They scroll past polished ads and bounce off landing pages.

AI traffic: Search queries and smart conversations

While traditional marketing channels buckle, AI-driven interfaces are quickly becoming the new front door to the internet. Tools like ChatGPT, Perplexity, and Copilot are reinventing the entire discovery journey. Instead of typing fragmented keywords into Google, people now initiate nuanced, context-rich conversations. A query like “best credit cards for travel” becomes a dialogue with an assistant that remembers what was asked five minutes ago, and tailors its answer accordingly.

What’s more striking is how users behave once they click through from AI responses. According to Adobe Analytics, traffic from AI sources consistently outperforms traditional traffic on all engagement fronts. Users spend more time on the site, navigate more pages, and bounce far less frequently. In concrete terms, visits from AI platforms show an 8% lift in time on site, a 12% boost in pages viewed per session, and a 23% drop in bounce rate compared to non-AI traffic.

That deeper engagement correlates with a dramatic rise in traffic volume. Adobe observed the first major inflection point during the 2024 holiday season. Between November and December, generative AI-driven traffic to retail sites spiked 1,300% year-over-year, with Cyber Monday alone jumping 1,950%. And the surge didn’t fade in January. As of February 2025, AI referral traffic was up 1,200% compared to just seven months prior, growing at a pace that’s doubling every two months.

This shift isn’t confined to retail. Banking and travel have seen parallel trajectories, with AI-driven visits growing 1,200% and 1,700% respectively in the same timeframe. Adobe’s data suggests that while AI referrals still make up a smaller slice of total traffic compared to channels like paid search or email, the growth curve is exponential and sustained.

The underlying reason lies in search intent. Semrush’s comparative analysis between ChatGPT and Google reveals a decisive tilt toward deeper, more exploratory user behavior in AI environments. On ChatGPT, over half of all queries,52.2%, are informational, far outpacing the 36.4% observed on Google. Navigational intent, which dominates Google at nearly 50%, drops to just over 34% on ChatGPT.

This means that users aren’t just looking for links, they’re trying to solve problems, understand concepts, and make informed decisions. AI isn’t replacing search in the way voice assistants tried to. It’s replacing the idea that search has to start with a link list.

And as this shift deepens, platforms are already moving to capitalize. ChatGPT now routes traffic with homepage-weighted links. Microsoft’s Copilot generates custom session journeys. Even analytics tools like Similarweb are building infrastructure to monitor AI-originated traffic.

AI isn’t just eating Google’s traffic; it’s reshaping how people discover, evaluate, and act. The brands that show up contextually inside these conversations, whether organically or through native AI ad formats, stand to win the next phase of digital discovery.

What happens next: Monetization inside the AI interface

Discovery has shifted. The old paths, SEO, email, influencer boosts, have lost their edge. Generative AI isn’t just a new traffic source; it’s an entirely different interface with different rules. And those who understand how it thinks, ranks, and routes will have a massive advantage.

But discovery is just the beginning. The next frontier is monetization inside the interface itself.

Platforms are already laying the groundwork. AI-native ad formats are emerging, embedded directly within conversations, APIs, and assistant workflows. The interface is becoming the inventory. In this world, attention doesn’t come in impressions; it comes in interaction windows, follow-up prompts, API calls, and co-piloted journeys.

Final thoughts: Marketing isn’t dead

What we’re witnessing isn’t the end of marketing. It’s the end of broadcast-era tactics pretending they still work in a world that’s moved on.

The old playbook — optimize content, buy reach, and push it everywhere was built for platforms that rewarded volume. But that era is closing fast. Saturation, declining performance, and hostile algorithms are symptoms of a deeper shift: attention can no longer be bought the same way.

What’s emerging in its place is a discovery system built on conversation, relevance, and intent. Generative AI is resetting the interface between brands and audiences and reprogramming how they evaluate, decide, and act.

In this new landscape, success will go to those who show up at the exact moment of need, embedded in the journey. The real opportunity is learning how to be discovered.

Because if the future of marketing lives inside a conversation, you’d better start learning how to speak the language.

The post Disrupting the MarTech stack: How generative AI is reshaping traditional marketing appeared first on Xenoss - AI and Data Software Development Company.

On July 17, Microsoft Advertising announced it will shut down its DSP, Microsoft Invest (formerly Xandr), by February 2026. Officially, the move is about aligning with “a more private and personalized advertising experience for a conversational and agentic world.”

Unofficially? It’s about shedding a clunky asset that never quite kept up.

Instead of keeping Invest on life support, Microsoft will now funnel its buy-side business into a streamlined, AI-assisted self-serve tool: the Microsoft Advertising Platform. Think less like DV360, more like a Performance Max clone with Copilot whispering in your ear.

Meanwhile, third-party DSPs will still be able to access Microsoft’s inventory. On the sell side, the company says it remains committed to Microsoft Monetize and Microsoft Curate—tools that serve publishers and control premium supply.

Invest’s lineage is notable: born as AppNexus, swallowed by AT&T, rebranded as Xandr, then offloaded to Microsoft in 2021. Now, it ends its run as another casualty of the AI consolidation wave.

Layoffs accompany the closure. The industry’s reaction? Mostly: “Saw it coming.”

The Xenoss take

Let’s not kid ourselves—this wasn’t just about “agentic futures” or a philosophical shift in UI metaphors. Microsoft’s decision to kill off Invest is, first and foremost, a cleanup act.

The DSP never truly recovered from the AT&T years. AppNexus was once a kingmaker. But after years of underinvestment and strategic whiplash, it fell behind. Microsoft gave it a new lease on life—but not a new direction. This was the slowest car on a crowded freeway, and eventually, someone had to pull over.

So they did.

But what’s interesting isn’t what’s dying—it’s what’s being built in its place.

AI eats the DSP

Microsoft is going all-in on first-party demand tools—faster, cheaper, and easier to control. AI is the new buying assistant. Not in the “help me choose a CPM” sense, but in the “I’ll take care of everything” sense.

If that sounds familiar, it should. Google’s Performance Max, Meta’s Advantage+, and Amazon’s growing suite of “just trust us” tools have been singing that tune for years. The selling point is simple: better outcomes, less effort. The cost? Visibility, control, and any illusion of a neutral marketplace.

Microsoft is just the latest to lean in. Invest’s shutoff clears the deck for AI-powered, walled-garden media buying where Microsoft controls the knobs—and the margins.

A shift in power, not philosophy

The blog post talks a big game about privacy and personalization, but let’s call it what it is: a business model pivot.

The money’s not in being a neutral DSP. The money’s in owning demand, supply, and the data pipes in between. Especially if you’ve got LinkedIn–level identity data to play with.

And therein lies the rub. Microsoft can talk privacy all day, but LinkedIn is a logged-in behavioral treasure trove. It’s not that they’re backing away from data. They’re just walling it in.

Invest didn’t have a place in that strategy. It didn’t generate enough revenue, didn’t have a strong differentiator, and worst of all, didn’t fit the narrative of a future shaped by Copilot-style automation.

What this means for the rest of the market

In the short term, The Trade Desk, Yahoo, DV360, and others will scoop up the displaced enterprise clients. Expect a flurry of “we’ll make your transition easy” emails by Q4.

Longer term, though, the DSP category faces existential questions. If every major platform is shifting toward AI-native, closed-loop systems, where does that leave the open web?

The answer might be: in the same place display has been for years—fragmented, under-monetized, and losing attention to platforms that offer simpler paths to performance.

For publishers, a small silver lining

Microsoft isn’t walking away from the supply side. Curate and Monetize will still be in play. But let’s be honest—those tools now function less like open-market facilitators and more like wrappers for Microsoft’s broader data and demand strategy.

In this light, Invest wasn’t sunset because the DSP model is broken. It was sunset because Microsoft’s DSP wasn’t big enough, sticky enough, or profitable enough to justify the overhead.

Final thought

Microsoft didn’t kill Invest because DSPs are obsolete. It killed it because generic DSPs without differentiated data or scale have no future in a market sprinting toward AI-powered, vertically integrated stacks.

This is less about the DSP model failing and more about Microsoft’s specific DSP failing to compete. Invest never became essential to buyers. It didn’t offer exclusive data, couldn’t compete with the firepower of The Trade Desk or DV360, and wasn’t sticky like Amazon Ads or Meta’s Advantage+ tools.

In an era where control of identity, first-party data, and proprietary media environments is the new holy trinity, running a standalone DSP with none of those advantages is a losing game. Microsoft knows where its strengths lie: LinkedIn, Bing Search, Edge, Outlook. Logged-in users. Authenticated data. AI-powered ad automation. That’s where it can control the rails, extract more margin, and create performance without having to explain the sausage-making.

So no—DSPs aren’t dead. But the “me-too” DSP is. If you can’t bring exclusive demand, differentiated data, or a smarter way to buy, you’re just middle tech—and middle tech gets squeezed.

The post Microsoft sunsets Invest DSP, signals a new era of walled-garden AI advertising appeared first on Xenoss - AI and Data Software Development Company.

Instead, the company says users can continue to manage tracking preferences through the existing privacy settings while it “remains committed” to Privacy Sandbox APIs. 

The reversal arrives after five years of shifting timelines, regulatory headwinds on both sides of the Atlantic, and growing scepticism from publishers and advertisers who have already poured millions into cookieless workarounds.

What led to the decision: Google’s illegal monopoly verdict

Google’s U-turn comes just days after a U.S. federal judge found the company holds illegal monopolies across several links of the AdTech chain, putting additional pressure on the search giant’s advertising strategy. 

Google says it will appeal, but a full victory seems unlikely, as two separate U.S. courts have now labelled the company an illegal monopolist (the first was the search verdict in September 2024). 

Why it matters for cookies: Every regulatory or judicial swing at Google’s ad stack shrinks the room the company has to experiment. 

Keeping third-party cookies alive in Chrome buys Google time to fight the breakup case without simultaneously rebuilding its core ad infrastructure.

Lena Cohen, staff technologist at Electronic Frontier Foundation, pointed out that “Google’s decision to keep third-party cookies—when every major browser blocks them—shows how an ad-driven business model trumps user privacy. We need robust U.S. legislation so ad companies don’t set the rules.”

Meanwhile, industry observers note that sustained regulatory pressure in the US and the CMA’s ongoing investigation in the UK left Chrome’s cookie deprecation plan politically fragile. 

Mathieu Roche, CEO of ID5, one of the leading providers of addressable advertising solutions, told Digiday, “I don’t believe in coincidences at Google. It feels like a negotiation tactic with the DOJ to keep Chrome inside the tent”. 

The Xenoss take

Same playbook, different chapter. 

Google has once again chosen the least disruptive path for its commercial interests—keeping the status quo intact while it fights antitrust battles on multiple fronts. 

From a purely economic perspective, the decision is rational: third-party cookies underpin a multi-billion-dollar demand-side marketplace that still flows predominantly through Google pipes. 

Turning that tap off while regulators hover with break-up orders would have been an unnecessary source of budget strain and technical pressure to offer viable alternatives.

Yet the climb-down leaves the broader ecosystem holding an expensive bag of “privacy-ready” investment. 

How the AdTech ecosystem should react to Google’s decision

CFOs and organizational leaders

Over the past 36 months, agencies, publishers, and MarTech vendors have funnelled R&D budgets into contextual engines, ID graphs, clean rooms, and probabilistic modelling—all predicated on Chrome finally shutting the cookie door. 

In 2025, CFOs will likely reconsider first-party data spend and push vendors to prove incremental value on top of the newly resurrected third-party cookies. Types of solutions reliant on third-party data (e.g., traditional DMPs) may experience a revival in the near future. 

Advertisers

For advertisers, the immediate implication is getting tactical breathing room. 

Targeting, frequency capping, and cross-site attribution pipelines remain intact, so campaign workflows will continue uninterrupted. 

Yet, staying complacent is not the most intelligent strategic bet. 

Privacy regulations—particularly in the EU—continue to tighten. Safari and Firefox still block third-party cookies by default, and the walled-garden duopoly (Meta and Amazon) never relied on them in the first place. 

In such a diverse privacy landscape, future-facing brands should not strive to bring cookies back to the forefront of their data strategies and continue diversifying data strategies with first-party assets and durable IDs.

Publishers

Publishers find themselves in a bittersweet position. 

An AdTech and Sales Director at a major European publisher told Digiday, “The whole process has been like watching a car crash in slow motion — trial-and-error, trial-and-regulator-rejection, trial-and-defeat.”

On one hand, cookie continuity stabilises open-web yield in the short term. On the other hand, Google’s shift may slow the buy-side appetite for direct, first-party data partnerships that have been gaining momentum over the last 3-4 years. 

To avoid sliding back into dependency, premium publishers will likely double down on authenticated traffic, consent orchestration, and value-exchange UX elements that deliver upside regardless of Chrome’s mood swings.

AdTech vendors

From a tech-stack perspective, the decision may accelerate consolidation. 

Businesses built solely on replacing cookies now face an existential funding crunch. In contrast, platforms that embed privacy-adaptive infrastructure, such as edge storage, on-device cohorts, and server-side API pipes, are suddenly differentiated. 

At the same time, organization leaders at AdTech companies echo the general sentiment of Google’s cookie backpedaling being tone-deaf to the industry’s privacy-preserving trajectory. 

In a VideoWeek interview, Adam Schenkel, EVP Global Platform Strategy at GumGum, commented, “Doubling down on cookies sends the wrong signal. The internet doesn’t have a targeting problem; it has a trust problem.”

Scope3’s agentic model—which optimises at the impression level with or without legacy IDs—looks prescient here, even if its sustainability claims warrant scrutiny.

Bottom line

Google’s volte-face should not be taken for a full-on cookie renaissance. 

The structural forces reshaping digital advertising—regulation, platform power shifts, and AI-driven optimisation—still come into play. Industry experts continue to state that, for third-party cookies, “the writing is on the wall,” and Google’s decision is not reflective of broader industry trends. 

To stay ahead of privacy-preserving changes, savvy brands and publishers should find a balancing act between continuous modernization to build a diverse data strategy and wisely choosing their investments in alternative IDs and cookieless solutions now that the sense of urgency is no longer present.

The post In the news: Google Chrome continues using third-party cookies appeared first on Xenoss - AI and Data Software Development Company.

But those profits don’t come as easy because behind those numbers lies an invisible battle for more precise real-time bidding (RTB) in programmatic advertising. Marketplaces face the mounting pressure to deliver higher ROAS. But many still only support hardcode fixed bids or struggle with crippling latency in real-time bid processing due to outdated tech stacks. 

The better news? Investments in RTB optimization with machine learning algorithms for advertising can drive massive gains. As part of our client portfolio, Viant Technology leveraged deep learning to build their AI Bid Optimizer—delivering to their clients 100% gains in performance improvements and 24% to 40% cost reduction compared to manual bidding systems. Coty, a multinational beauty company, improved its ROAS from Amazon ads by 28% and its total digital purchase view rate by 48% through bid modifier API integration. 

More and more media buyers realize they can’t compete with ML-based bidding optimization algorithms in terms of speed or precision. So they now expect their demand-side partners to step up to the game with better tech (or risk being replaced).

However, many companies have been slow to respond because of the challenges associated with RTB bidding automation: high-dimensional data, ultra-fast bid response speeds, and a general lack of knowledge in implementing online ad optimization with machine learning. 

Does this sound like your case? In this post, you’ll discover: 

• The fundamental Fair Price bidding problem in RTB that complicates campaign optimization 

• RTB optimization challenges specific to machine learning like class imbalance, cold start problem, and concept drift 

• Proven solutions to both illustrated via case studies from top online advertising marketplaces 

Core challenges in RTB campaign optimization

Theoretically, RTB bidding campaign optimization is a straightforward prediction problem — estimate conversion, set a bid, win the impression. 

But on large online marketplaces, it’s anything but that. The underlying algorithms must adapt to skewed data, changing behaviors, and scale-driven complexity to generate each bid. 

Here’s a breakdown of the biggest technical challenges standing in the way of ML-powered programmatic bidding. 

Fair price bidding: A more innovative RTB strategy

At its core, real-time bidding (RTB) is an economic optimization problem. For every single ad impression, a programmatic RTB platform has to answer one deceptively simple question:

Given its potential to convert, what’s the maximum we should pay for this impression?

This is the “Fair Price” problem. And many real-time bidding algorithms fail to solve it in probabilistic environments like retail media networks. 

Let’s say you’re running a retargeting eCommerce advertising campaign for a popular pair of sneakers. An impression becomes available for a user who:

• Visited the product page for the sneakers 2 hours ago
• Added the item to the cart but didn’t check out
• Bought a similar item about a month ago 

Now, here’s the question: How much should your DSP bid for this impression? 

Machine learning programmatic advertising algorithms will dynamically estimate the probabilistic value of every impression based on two factors:

• MaxCPC is the advertiser’s maximum willingness to pay for a click 
• PredictedCTR is the estimated probability that this impression will result in a click

Knowing these two parameters, you can calculate the Fair Price using this formula: 

Let’s assume your max CPC is $5 (based on ROAS targets). The underlying machine learning advertising model estimated a CTR of 5% based on user behavior signals. In this case, the best bid would be $0.25 per impression. 

By using an ML-powered programmatic bidding strategy, you can avoid the pitfalls of fixed bigging strategies like: 

• Overpriced bad impressions. Not all ad impressions are created equal. A user who bounces in 0.5 seconds with no scroll activity has zero conversion potential. A fixed $0.50 bid on that impression? You’re overpaying for junk inventory.
• Lost valuable impressions. On the flip side, a high-intent user, returning visitor, product page view, or high basket value might only convert if you outbid competitors. A fixed bid that’s too low (say, $0.10) means you lose the auction and a conversion opportunity.
• No context, no efficiency. Fixed bids don’t account for device type, time of day, user recency, creative performance, or even page category — all of which meaningfully impact conversion probability.

The “Fair Price” model is a game changer for preventing marketplace ad waste. But implementing it in real time and at scale opens up a new can of worms.

ML challenges in implementing fair price bidding

The fair price formula provides a clear theoretical framework for optimizing RTB programmatic buying. However, implementing it in an AdTech platform is far from trivial. 

Accurate bidding depends entirely on the quality of the underlying prediction models — and building those models at scale introduces significant ML and data engineering complexity.

The class imbalance problem

Conversion probability predictions are complicated because the positive class (clicks and conversions) is vanishingly rare. The average CTR for retail media advertising is just 0.39%. Amazon marketplace advertising, in turn, averages 0.42%. The conversion rates can be even lower, depending on product category, funnel depth, and user behavior. 

Respectively, machine learning algorithms for advertising are mostly trained on highly imbalanced datasets, which causes two problems: 

• Overprediction of the majority class. The algorithm defaults to labeling most ad impressions as non-converting because that’s statistically the safest bet in an imbalanced dataset.
• Suppression of the minority class. To minimize false positives, the model becomes overly conservative and fails to identify genuine conversion signals. 

In other words, a machine learning media buying model never learns to recognize the right patterns despite showing a high AUC or accuracy score on a monitoring dashboard. 

The business results? Underbidding on high-value impressions leading to lost revenue and
overbidding on low-propensity users (aka wasted ad budgets), multiplied a billion with every subsequent RTB bid.  

High-dimensional data

To accurately value each impression, machine learning advertising models must take into account a heap of variables about the user and the product, plus the merchant, ad format, device, time of day, and real-time behavioural context. 

In an average retail media network, that’s millions of potential features per impression. Most of these are categorical variables with high cardinality, meaning they don’t play nicely with many standard ML models.

The following problems emerge: 

• Too large feature marcices.  When encoding high-cardinality variables (e.g., 100K+ user IDs in one go), the feature matrix becomes too big for efficient training and storage.
• Complex data dependencies. The trifecta of user-merchant-product interactions creates different CTR and conversion probabilities, which also change depending on the pricing, season, or user demographics. 

• Challenging dimensionality reduction. Naive approaches can discard high-impact features from the model or introduce bias, leading to skewed performance. 

• High resource requirements. Advanced RTB programmatic advertising models need feature store architectures that serve data in <10ms latency, meaning high cloud infrastructure costs and complex GPU memory optimization. 

To overcome these problems, data science teams need to compress categorical variables (e.g., user ID, product ID) with the right embedding techniques into low-dimensional dense vectors. 

Marketplaces like Alibaba use trainable embedding tables to optimize high-cardinality entities. These are dense vector representations, optimized via back propagation during the model training phase and then embedded into neural models to support CTR prediction. 

Another strategy is to sparse feature engineering pipelines with feature hashing or target encoding. This way, you can avoid maintaining up-to-date embedding vectors for every possible user or product. Instead, the model fetches learned embeddings for frequent entities or hashed representations for rare or unseen ones, improving the speed of programmatic bids. 

Real-time constraints

In RTB, everything — including the fair price calculation — has to happen in real time. So, model speed is paramount. 

Every impression triggers a programmatic auction. Every DSP has a hard deadline (a 100ms timeout per IAB OpenRTB 2.x specification) to respond to a bid request. Miss that window? Your bid has dropped.

During this small window, the real-time bidding programmatic model must also compute an optimized bid: 

• Predict the probability of a click or conversion
• Analyze the current state of campaign pacing and budget
• Apply frequency capping, brand safety, merchant priorities, etc 

This is where it gets tricky: More complex models — like deep neural networks with embedding layers and cross-feature interactions — generally offer better predictive bid performance. But they’re also slower to compute, especially under live traffic.

So, every online advertising marketplace has to consider tradeoffs between model accuracy and tolerable latency levels. AdTech engineers must make hard decisions: 

• Do you serve precomputed predictions?
• Do you trim down features at inference time?
• Can you cache predictions for certain user segments?
• Should you use a faster approximation model (e.g., tree-based or linear) instead of deep learning?

These questions become more acute as the platform volume increases. The biggest marketplaces must make tens of millions of predictions per second — each of which needs to be completed in under 10ms, including network round-trip time, feature lookup, model inference, and response generation. When your model is too slow, you lose revenue opportunities. 

Pro tip from our MLOps team: Try the XGBoost algorithm

XGBoost models are highly optimized for low-latency predictions — ideal for sub-millisecond scoring per impression. They also bode well with sparse inputs natively without needing heavy preprocessing. Compared to deep learning, XGBoost models are “lightweight”, making them ideal for edge deployment or in-memory scoring at scale in RTB.

Cold start problem

When an ad bidding model doesn’t have enough historical data for CTR and conversion probability, we have a cold start problem. 

The Predicted CTR term in the Fair Price bidding formula becomes unreliable (or impossible) to calculate when the processed impression involves: 

• A new merchant with no past campaign performance
• A new product without browsing or engagement history
• A new campaign targeting novel audiences or ad placements

In such cases, the model will either produce a flat, uninformed CTR (leading to underbidding or overbidding) or not bid at all — missing impressions that may have performed well. 

The problem gets worse at marketplace advertising because inventory constantly changes, new merchants get onboarded frequently, and short-term campaigns don’t produce enough training data. All of this makes cold start the default state for large portions of traffic.

To overcome this shortcoming, some AI programmatic advertising platforms use meta-learning techniques to transfer knowledge from past campaigns. 

WeChat, for example, developed a two-stage framework called Meta Hybrid Experts and Critics (MetaHeac) to enhance audience expansion in their advertising system. The company first trained a general model using data from existing advertising campaigns to capture the relationships among various tasks from a meta-learning perspective. Then, for each new campaign, a customized model is derived from the general model using the provided seed set. This approach allows the system to adapt quickly to new campaigns by leveraging insights gained from previous ones.​ 

Another approach is to create similarity-based models to effectively infer the optimal CTR from similar products, merchants, or customer segments. A recent paper showcases a Graph Meta Embedding (GME) model that can use two information sources (the new ad and existing old ads) to improve the prediction performance in both cold-start (i.e., no training data is available) and warm-up (i.e., a small number of training samples are collected) scenarios on different deep learning-based CTR prediction models. 

The key to solving the cold start problem is striking a balance between exploration and exploitation. Over-exploit, and your system stagnates. Over-explore, and you waste money.

Effective programmatic ad platforms dynamically balance both — adjusting exploration rates based on budget pacing, merchant goals, and model confidence scores.

Concept drift and model decay

User behaviors, product demand, and competitive bidding patterns constantly change. That means a model calibrated for yesterday’s data is already drifting off-course today. 

Fair Price model drift happens due to: 

• User behavior shifts. Customers’ interest changes (no longer into skincare) or their device usage patterns evolve (e.g., they start shopping more from an iPad).
• Marketplace dynamics. New brands, SKUs, and categories introduce nonstationary data, while supply/demand fluctuations (e.g., stockouts) affect impression volume and pricing.

• Seasonality effects. For many products, conversion rates aren’t the same throughout the year. Holidays (e.g., Christmas, Thanksgiving, and Black Friday) also alter CTR baselines.
• Competitive pressures. Aggressive budget increases from other advertisers distort the auction landscape. Also, new brands may cannibalize your audience segments. 

All of these change the underlying relationships that your model learned — between user intent, product type, and likelihood of conversion. This is known as concept drift — and it’s inevitable in programmatic RTB systems. 

To counter model degradation, you have to implement continuous monitoring and retraining workflows with: 

Some platforms also use more advanced techniques to prevent machine learning drift. Pinterest has integrated a Multi-Task Learning (MTL) technique to optimize its ads conversion models.  MTL combines multiple conversion objectives into a unified model, leveraging abundant onsite actions to enhance training for sparse conversion objectives. 

This approach allows the model to adapt to new data incrementally, reducing the odds of drifting. Additionally, Pinterest employs ensemble models that combine different feature interaction modules, such as DCNv2 and transformers, to capture diverse data patterns effectively and use them in programmatic campaigns. 

Solution: The strategic advantage of advanced RTB optimization

Building an effective machine learning programmatic advertising system for a marketplace can feel like a minefield of edge cases, system constraints, and modeling tradeoffs.

But effective solutions exist — and our programmatic software development team has successfully implemented them. By applying systematic, end-to-end optimization strategies for our client’s platform, we have achieved double-digit lifts in CPC and CTR, along with a massive reduction in operating costs.

• 27% CPC reduction compared to initial rule-based bidding strategies
• 18% CTR lift and 9% CR lift through better identification of high-intent audiences
• 45% reduction in operating costs, thanks to a fully automated model maintenance cycle that significantly reduced the number of campaign management tasks

Here are two approaches we used to achieve these results: 

Multi-model architectures

In a large-scale marketplace, using one machine learning model to optimize bids never works. Different product categories attract fundamentally different user behaviors, intent signals, and conversion patterns — and one-size-fits-all modeling fails to capture this complexity.

That’s why we encouraged our client, one of the largest retail marketplaces in CEE, to move to a multi-model architecture. A multi-modal AI optimization system classifies products into behavioral clusters (e.g., by category taxonomy or brand patterns) and trains separate models for each group.

This approach effectively addresses class imbalance. Each model gets trained on homogeneous data subsets, meaning a more balanced signal-to-noise ratio. It also reduces high-dimensional data complexity as feature distributions become narrower and more predictable within a specific category. 

Lastly, such models can be fine-tuned to category-specific KPIs, latency tradeoffs, or budget pacing rules, offering greater future-proofing. 

Automated model retraining pipelines

The sheer volume of changing data points renders manual model retraining absolutely ineffective. That’s why modern RTB platforms adopt self-healing ML systems. Automated pipelines:

• Monitor model performance in real time (e.g., AUC, CTR lift, cost per action)

• Automatically detect degradation due to drift or data distribution changes.
• Trigger on-demand retraining or model replacement using recent data windows or pre-trained category-specific baselines.

This approach doesn’t just mitigate concept drift — it also helps solve the cold start problem. When a new campaign launches, the system can pull from a repository of pre-trained models, fine-tune on limited data, and bootstrap the learning process autonomously.

In the case of our client, automated model retraining pipelines enabled:

• Gradual model rollouts with traffic splitting and live A/B testing
• Performance-based decisions for model swaps — not just on schedule but based on actual campaign-level KPIs
• Zero manual ops even across thousands of active models, leading to greater operational efficiencies

Ultimately, automated retraining turns RTB modeling from a brittle, manual process into a robust feedback loop — where models continuously self-adapt, learn, and optimize. 

Conclusion 

RTB bidding for marketplaces isn’t just another machine learning challenge — it’s a multi-dimensional problem that spans economics, system design, model engineering, and real-time constraints. The complexity is undeniable, from predicting fair prices to handling concept drift, cold starts, and massive feature spaces.

But it’s also solvable. The difference between theoretical success and real-world results lies in the implementation details — how you structure your model architectures, monitor performance, automate retraining, and scale decision-making across thousands of campaigns. This is where practical ML expertise and system-level thinking make all the difference.

To get a deeper scoop, check our detailed case study showcasing how a leading European marketplace with 128,000 merchants implemented this approach to achieve remarkable results — including a 27% reduction in CPC, an 18% CTR lift, and a 45% reduction in operational cost.

The post RTB optimization for marketplaces: Solving programmatic complexity with machine learning appeared first on Xenoss - AI and Data Software Development Company.