Table of Contents
The new whitepaper from Realize, How to Close the Data Fragmentation Gap on the Open Web, looks at how advertisers and agencies can build a more continuous view of the people they’re trying to reach outside the confines of walled gardens.
The approach starts with Realize ID. This connects identifiers such as cookie-based IDs, mobile advertising IDs, publisher IDs, third-party IDs, and RampID into one profile. The Realize Identity Graph processes 240–390 million distinct clusters each day to keep those profiles current.
Recognition only gets you so far, though. Knowing that the same person has come back is useful, but the bigger picture comes from what they do next: what they read, what they search for, what they engage with, where they drop off, and whether they eventually convert.
Because Realize’s code runs across publisher pages and the Taboola Pixel sits on advertiser sites, you can see more of that journey as it unfolds. Those signals help you to understand how interest develops, then feed what happens later back into the system so future predictions can improve.
This research brings together external industry studies, our own data, and advertiser results. Together, they point to the same idea: solving fragmentation takes more than matching identifiers. You need a way to recognize the same person over time, understand the signals they leave behind, and connect those signals to real campaign outcomes.
Download the Realize ID Whitepaper
Key Takeaways
1. Identity Resolution Is the Starting Point
Most of the open web isn’t logged in. In 2025, 84% of publishers used or planned to use identity graphs, per the whitepaper, while 91% were testing alternative IDs. Yet, publishers with even 60% authenticated audiences remain uncommon.
Realize ID works across that fragmented environment by clustering the identifiers a user leaves behind into one profile. Instead of treating every new device, browser, or domain as a completely new person, Realize ID can maintain a more persistent identity across the journey.
2. The Signals Attached to an ID Determine How Useful It Becomes
Knowing that a user returned doesn’t tell you what they want. Search activity, reading behavior, ad engagement, page context, leads, carts, and purchases add the context needed to understand intent. Those signals help move targeting beyond a static description of who someone resembles, toward what they may be considering now.
3. Closing the Loop Improves More Than Targeting
Once impressions, engagement, conversion activity, and first-party data can connect to the same identity, that profile can support bidding, frequency capping, retargeting, measurement, and audience activation. Every new outcome also feeds the next decision. The profile that informed yesterday’s bid can become more useful after seeing what that user did next.
Introduction
Why Data Fragmentation Has Become an Open-Web Problem
People spend their day across publishers, apps, email, search, commerce sites, social platforms, and, increasingly, LLMs. US adults now spend nearly 13 hours a day with various types of media, and a purchase decision can move through several of those environments before it reaches an advertiser.
Walled gardens have an advantage when the entire journey happens inside a logged-in ecosystem, as identity, engagement, and conversion can all sit on the same infrastructure. While the open web offers a much broader view of a person’s day, historically, that view has been fragmented.
What This Research Covers
This research examines what it takes to connect that journey, such as resolving multiple identifiers into one identity, attaching useful behavioral signals, enriching it with advertiser data, and feeding campaign outcomes back into future decisions.
The Open Web’s Data Fragmentation Problem
Every player in the advertising ecosystem sees a different part of the customer journey, from bid and click data to inventory, publisher engagement, and first-party customer records. The difficulty is connecting those signals well enough to recognize that they belong to the same person. That creates practical problems: an advertiser may pay to reach one buyer several times under different identifiers; a returning visitor may be treated as a new prospect; earlier impressions can become disconnected from a later conversion.
Adding another identifier doesn’t automatically fix the problem. What the open web needs is infrastructure that connects identities across environments and carries useful information.
How Realize ID Connects the Journey
Connect Multiple Identifiers Into One Profile
As people move across the network, they leave behind different identifiers. These can include cookie-based IDs, mobile advertising IDs, publisher IDs, third-party IDs, hashed identifiers, and RampID. Realize ID clusters all those identifiers into one profile. The Realize Identity Graph processes 240–390 million of these distinct clusters every day, continually refreshing the connections between them.
That gives us the identity layer. The next question is what we can learn once we’ve recognized the user.
Attach Signals That Reveal Intent
Signals record what users do. They can include the searches they make, the content they read, how they engage with a page, which ads they interact with, what they add to a cart, and whether they eventually convert.
The quality of those signals varies considerably. In a 2025 study from CIMM, Go Addressable, and Truthset, purchased identity linkages from six major providers were accurate just 16% of the time — or less. The providers agreed on fewer than 7% of the linkages.
Realize ID collects signals across the places where Realize is active, following how interest develops over time. On publisher pages, we can see how people engage with content. Once they reach an advertiser’s site, the Taboola Pixel picks up what happens next, whether that’s a lead, a checkout, or a purchase. Those signals keep feeding back into the system: We learn from what people do, use those patterns to improve targeting and predictions, then measure the outcome and refine the next decision.
Add What You Already Know About Your Customers
Advertisers can add another layer with their own first-party data. That might include high-value customers in a CRM, lapsed or loyalty segments in a CDP, or even an agency’s own identity graph.
Realize can match that data through hashed first-party uploads, DMP or CDP connections, custom ID syncing, and clean-room matching. Once we make the match, we can combine what the advertiser already knows about those customers with the live behavior we see across the network. That gives an existing audience more current context, including signals that can show what people are interested in now, and how close they may be to taking action.
Evidence and Research
Realize internal data from this year shows how that combination can affect performance. Signal-enriched predicted intent models have produced up to a 144% uplift in conversion rate. Users reached through our 3,000+ In Market For segments convert at rates 20% higher, while search and mail-domain signals have produced CPA reductions of up to 40% and 28% respectively.
These results vary by advertiser, campaign, market, and setup, but the wider pattern is that the closer audience selection gets to observed intent, the more useful identity becomes as a performance signal.
Applications and Use Cases
Livguard Finds Higher-Intent Traffic Beyond Traditional Channels
Livguard wanted traffic that showed deeper consideration, not simply more visits. The brand used Realize ID alongside contextual targeting, behavioral audiences, and sequential retargeting to reach people showing interest in energy and power-backup products. Realize traffic produced 16% longer session duration than paid search and 29% more actions per session than social.
Mercedes-Benz and OMD Taiwan Improve Session Quality
Mercedes-Benz Taiwan and OMD Taiwan used Realize ID to reach people showing luxury automotive intent across native and display campaigns. Across its sales campaigns, Realize delivered up to 63% higher engagement rates, nearly 2x the time on site, and the lowest cost per session across its media mix. Mercedes-Benz has since made Realize part of its ongoing media planning in Taiwan.
Philips Home Appliances Turns Audience Signals Into Conversions
Philips Home Appliances used Realize ID to reach high-quality audiences, combined with Motion Ads and Maximize Conversions, to promote its Garment Steamer.. The campaign delivered a 25% higher conversion rate and a 30% lower bounce rate than the network average, along with more qualified leads and orders.
Recommendations for Advertisers
Look Beyond Match Rates When Comparing Identity Solutions
Recognition is only one part of an identity graph. Ask where the signals attached to that identity come from, how current they are, which parts of the customer journey the provider can observe, and whether outcomes can feed back into future campaign decisions. Remember, a large graph with weak signals can still leave you guessing about intent.
Connect First-Party Customer Knowledge With Current Behavior
Your CRM or CDP may already tell you which customers are valuable. What it may not tell you is what those customers are interested in today. Combining your own customer context with current behavioral signals can create more useful audiences for targeting, suppression, re-engagement, bidding, and measurement.
Conclusion
The open web gives advertisers access to a rich and varied customer journey. Its weakness has been continuity. Realize ID is built to close that gap by connecting identifiers, adding signals that reveal intent, incorporating advertiser data, and linking downstream outcomes back to the identity that started the journey. This approach gives advertisers a stronger foundation to reach people based on where they are in the buying decision, measure what happens afterward, and feed those outcomes into the next campaign decision.
Frequently Asked Questions
Why is so much programmatic spend wasted on the open web, and what causes it?
The ANA estimated $26.8 billion in wasted programmatic spend in 2025, and fragmented identity is one part of that wider programmatic efficiency problem. When the same person appears under several identifiers and impression-level data is difficult to connect, advertisers can duplicate reach and lose visibility across the journey.
Every vendor has an identity graph now. How do I tell them apart?
Look at what gets attached to the identity after the match. Signal provenance, recency, first-party data connectivity, activation options, and access to downstream outcomes can tell you more than graph size alone.
How can I recognize the same user across devices and domains when most of the open web isn’t logged in?
Realize ID clusters multiple identifiers associated with the same user into one profile, including cookie-based IDs, mobile advertising IDs, publisher IDs, third-party IDs, and RampID.
Can I get frequency capping and closed-loop attribution outside walled gardens?
Yes. A frequency cap tied to a Realize ID can apply to the user rather than each identifier they carry. When that person converts, activity recorded through the Taboola Pixel can also connect the outcome back to earlier campaign interactions.
How do I activate my CRM or CDP data without relying entirely on a black-box clean room?
You can connect first-party data through hashed file uploads, supported DMP or CDP integrations, custom ID syncing, or clean-room matching. Match rate and audience size are visible in Realize, giving you more insight into the audience before activation.
Does intent-based targeting outperform demographic targeting on lower-funnel outcomes?
Our data shows strong results from intent-led signals. Users in In Market For segments convert at rates 20% higher, while search keyword and mail-domain signals have produced CPA improvements of up to 40% and 28%, respectively. Performance will vary by campaign, but these results show what becomes possible when targeting reflects current behavior rather than demographics alone.
Research Methodology
Research Objective
We examined why identity remains fragmented across the open web and how connecting identity, behavioral signals, advertiser data, and outcomes can create a more continuous view of the purchase journey.
Research Approach
The research combines published industry studies, Realize platform data, and advertiser campaign results.
Data Sources
External sources include the ANA, eMarketer, ID5, CIMM, Go Addressable, Truthset, and Skai. We also used Taboola internal data and advertiser campaign results.
Data Collection and Analysis
We drew industry statistics from published research cited throughout the paper. We drew platform performance findings from our internal campaign and identity data, while use cases show how individual advertisers have applied Realize ID.
Scope and Inclusion Criteria
The research focuses on identity resolution, intent signals, first-party data activation, audience targeting, measurement, and campaign performance across the open web.
Limitations
Internal performance results and advertiser case studies reflect specific campaigns, audiences, markets, and configurations. They shouldn’t be interpreted as guaranteed results for every advertiser.
Research Date
The research was produced in 2026 and draws primarily on industry studies published in 2025 and 2026.
Sources and References
Research referenced in the whitepaper includes:
- ANA, Programmatic Transparency Benchmark (2025)
- eMarketer, US Time Spent With Media (2026)
- ID5, State of Digital Identity (2025)
- CIMM, Go Addressable, and Truthset, Challenges of Using IP Address for Audience Identification and Measurement (2025)
- Skai, State of Retail Media (2025)
- Taboola internal data (2026)
- Advertiser campaign and case-study data
About the Authors
The Realize ID whitepaper was compiled by the following Taboola SMEs:
Content: Amanda Gillette, senior marketing manager.
Product marketing: Katherine Pickles, product marketing director.
Product: Tomer Tunitsky, senior product manager.
About Taboola
Taboola empowers businesses to grow through performance advertising technology that goes beyond search and social and delivers measurable outcomes at scale.
Thousands of businesses advertise directly through Realize, our performance advertising platform, reaching approximately 600 million daily active users across some of the world’s leading publishers.
Publishers including NBC News and Yahoo, and OEMs including Samsung and Xiaomi, use our technology to grow audience and revenue, giving Realize access to unique data, specialized algorithms, and broad scale.
