Identity Resolution: How to Solve Fragmented User Identity

A single buyer can look like several different people to an advertising platform. They might research a product on their phone, compare options on a laptop, and return later through a different browser, generating discrete identifiers at each step.

Identity resolution solves that fragmentation by determining which identifiers belong to the same person and connecting them into a persistent record. Instead of treating every device, browser, session, or interaction as an unrelated user, identity resolution allows platforms to build a more consistent view of the person behind them.

Importantly, identity resolution isn’t a one-time action or a feature that simply gets switched on, it’s a continuous process. New identifiers appear, old ones expire, user behavior changes, and connections must be continually evaluated and updated.

For performance marketers, fragmented identity affects everything from reach and frequency to targeting, suppression, and measurement. As customer journeys span more environments and touchpoints, understanding how identity resolution works and where its limitations lie is critical.

What Is Identity Resolution?

Identity resolution is the process of determining that multiple identifiers observed across devices, browsers, and environments belong to the same person, and connecting them into a single persistent record.

People generate multiple identifiers as they move between devices, browsers, websites, apps, and other digital environments. Without identity resolution, those identifiers can remain disconnected, which causes a single person to appear as multiple unrelated users. Resolution attempts to establish connections between those identifiers. When sufficient evidence indicates that multiple identifiers represent the same person, you can combine them into one persistent record. That record might not establish who someone is by name, but it can create a consistent representation of the same user across the identifiers connected to them.

Identity resolution is also continuous. Identifiers appear, change, and expire, so relationships between them must be maintained and updated over time. A platform may also gain additional signals that strengthen or weaken existing connections.

For this reason, you should never view identity resolution as a process that perfectly resolves every user. Platforms resolve more identities as they gather and connect additional information, while the quality and coverage of those connections can vary.

What Identity Resolution Is Not

Identity resolution overlaps with several concepts in advertising technology, but they aren’t interchangeable. Avoid confusing identity resolution with the following:

It is not an identity graph. Identity resolution is the process; an identity graph is the structure it reads from and writes to. The graph links users to their identifiers across devices, browsers, and websites.

It is not tracking. Tracking collects observations about user activity. Identity resolution connects observations and identifiers that a platform already has.

It is not a single vendor identifier. Shared identifiers can be useful inputs into identity resolution, but an individual identifier is not the resolution process itself.

It is not de-anonymization. Identity resolution does not establish a person’s real-world identity or determine their name. It determines whether different identifiers are likely to represent the same person.

How Identity Resolution Works, Step by Step

Implementations differ between platforms, but identity resolution generally follows the same basic process: collect identifiers and signals, standardize them, identify potential matches, cluster connected identifiers, and continuously maintain those relationships. Here’s a closer look at how each step works:

Collecting Identifiers and Signals

Platforms receive identifiers and signals from the environments where they operate. Identifiers help distinguish users, while signals provide information about their behavior, interests, context, and interactions.

Normalization

To reliably compare identifiers, incoming data needs to be standardized. Differences in formatting, structure, or processing can prevent records that should match from being recognized as related. Normalization puts comparable data into consistent formats so the matching process has a reliable foundation.

Matching

Once data is normalized, the platform evaluates whether different identifiers are likely to belong to the same person. Matching methods can carry different levels of confidence, so platforms must evaluate the strength of potential connections, rather than simply assuming that similar observations belong together.

Clustering Into a Single Record

After connections have been established, related identifiers can be grouped into a single record, or cluster. This is the point at which previously fragmented observations begin to represent a single user, rather than several disconnected identities. The resulting unified ID provides a persistent reference for the cluster.

Maintaining and Updating

Identity resolution doesn’t end once a cluster is created. New identifiers will continue to appear, while others expire or become less useful. Additional observations can strengthen existing connections or change how a record should be interpreted. Ongoing maintenance is essential because identity resolution decays over time. An identity graph is therefore never truly finished; it must continuously adapt as the underlying data changes.

Identity Resolution at a Glance

What Is a Unified ID?

A unified ID is the single persistent record that identity resolution produces, representing one person across all identifiers associated with them. It’s also commonly described as a cluster, because it groups multiple connected identifiers around the platform’s best representation of a single user.

This is different from an individual identifier, which might represent a particular browser, device, login, or other observable reference point. A unified ID sits above all individual identifiers and consolidates them into a single record; it provides platforms with a more stable reference point as users move between environments.

Individual identifiers may appear or disappear, but the unified record can persist as long as the platform has enough information to maintain the connection.

How a Unified ID Gets Created

A unified ID is the output of the clustering stage of identity resolution. After identifiers have been collected, normalized, and matched, identifiers believed to represent the same user are grouped into a cluster. The unified ID becomes the persistent reference for that cluster.

Persistence is what makes the unified ID particularly useful. If one identifier expires, the entire user record does not necessarily disappear. Other connected identifiers can continue to support the cluster, and new identifiers can be connected as they become available.

Realize ID: Our Implementation

At the core of Realize is Realize ID, Taboola’s unified ID that connects fragmented identifiers into a single, persistent view of a user. It can cluster more than 140 connected identifiers, including hashed emails (HEMs), mobile advertising IDs (MAIDs), RampIDs, and publisher partner IDs (for a list of useful terms, visit our Realize ID Glossary page). Realize associates signals with these clusters to build a more complete user profile across fragmented touchpoints.

What Resolution Changes in a Campaign

Identity resolution is important because fragmented identity creates practical problems for campaign execution. By connecting identifiers, advertisers can better understand reach, control exposure, suppress audiences, measure outcomes, and interpret behavioral signals.

Reach and Frequency You Can Trust

If one person appears as three different users, campaign reach can be overstated, and frequency understated. Resolution allows connected identifiers to be treated as one user, which gives advertisers a more accurate view of how many people they reached, and how often those people were exposed to advertising.

Suppression That Works

Fragmentation can also cause advertisers to keep targeting people after they’ve already completed the desired action. Connecting a conversion to the same persistent record used for targeting makes suppression more effective. Instead of suppressing only one identifier, the campaign can account for the other connected identifiers representing that user.

Measurement That Connects Exposure to Outcome

Resolution can help connect advertising exposure and subsequent outcomes that occur through different identifiers or environments. This provides measurement systems with a more complete basis for understanding whether interactions at different points in a customer journey belong to the same user.

Targeting a Buyer Rather Than a Fragment

A person’s behavior rarely occurs on just one device or during one session. Someone might read about a product category on one device, compare options on another, and later engage with an ad in a third browser. When those observations remain disconnected, each represents only part of the buyer’s behavior. Identity resolution connects those fragments, allowing the signals associated with them to contribute to a more complete user profile.

What Resolution Looks Like at Different Scales

At its core, the identity resolution process doesn’t change with advertiser size. What does change is the amount and complexity of first-party data being resolved, and how the advertiser uses it.

Smaller Advertisers

Smaller advertisers may have relatively modest customer lists, but identity resolution can still be very useful. In fact, a smaller, first-party footprint can make the platform’s broader footprint in performing resolution particularly important. Connecting available first-party data with identifiers observed across a wider network can help advertisers make more effective use of the audience data they already have.

Enterprise Advertisers

For enterprises, the challenge is often fragmentation inside the organization itself. Customer information may be distributed across CRM platforms, CDPs, loyalty programs, offline systems, and other databases that were not designed to work together. Identity resolution can connect those observations into a single record before that information is used for media activation.

Agencies

Agencies face the challenge of operating campaigns across many advertisers, each with its own data assets and identity infrastructure. Resolution doesn’t necessarily replace an agency’s existing identity graph. Instead, an agency’s graph can serve as input, allowing existing identity intelligence to work alongside the advertising platform’s resolution capabilities.

Where Identity Resolution Reaches Its Limits

Identity resolution can reduce fragmentation, but it cannot eliminate it.

First, resolution is limited by footprint. A platform can only resolve identity across the properties, data sources, and environments where it has the relationships and signals necessary to establish connections. Activity outside that footprint may remain invisible. That makes footprint one of the most useful factors for comparing identity resolution providers: the broader and more relevant the observable environment, the greater the opportunity to connect interactions that are otherwise fragmented.

Second, authenticated traffic impacts the availability of high-confidence matches. Logged-in environments can provide persistent identifiers that create strong connections, while unauthenticated environments may offer fewer opportunities to establish those relationships.

Finally, confidence isn’t uniform across an identity graph. Some connections may be supported by strong evidence, while others carry less certainty. Identity resolution therefore isn’t simply a question of whether a user has been “resolved;” the quality of the connections matters, too.

These limitations are another reason identity resolution must be treated as an ongoing process, rather than a finished database.

How to Evaluate a Platform’s Resolution Capability

When comparing the capabilities of identity resolution, advertisers and agencies should look beyond whether a platform simply offers identity resolution. The more useful questions examine how that resolution actually works:

• Across what footprint is identity resolution performed? Consider where the platform can observe and connect identifiers and where gaps remain.
• What share of its reach is authenticated? Authentication can affect the availability of persistent, high-confidence identity connections.
• Is identity resolution applied to measurement as well as targeting? Connecting users for audience activation is only one potential application.
• What happens when an identifier expires? A useful identity system needs to maintain records as individual identifiers change or disappear.
• Can the advertiser’s own first-party data be resolved into the same record? First-party data becomes more useful when it can contribute to the same persistent user view used for campaign activation.

Key Takeaways

Identity resolution is ongoing; it’s not a single feature that simply turns fragmented data into perfect user profiles. It determines when identifiers across devices, browsers, and environments represent the same person and connects them into persistent records. Those records can improve reach and frequency accuracy, audience suppression, targeting, and measurement.

That said, resolution is only as useful as the data and relationships supporting it. Platform footprint, authentication, maintenance, and connection confidence all affect what can be resolved. For marketers, the goal isn’t simply to find a platform that “has identity.” It’s to understand how effectively that platform can connect the fragmented customer journeys that matter to their campaigns.

Frequently Asked Questions (FAQs)

Is identity resolution the same as an identity graph?

No. Identity resolution is the process; an identity graph is the structure it reads from and writes to. Resolution determines which identifiers belong to the same person and connects them. The identity graph stores and maintains those relationships across users, identifiers, devices, browsers, and websites.

Is identity resolution privacy-safe?

Identity resolution can be designed around privacy-conscious practices, including consent management, first-party data, and privacy-safe identifiers. However, identity resolution itself does not guarantee privacy or regulatory compliance. Advertisers and platforms still need appropriate data governance and consent practices for the jurisdictions in which they operate.

Does identity resolution require login?

No. Authentication can provide strong, persistent identifiers that improve matching confidence, but not every identity connection requires a login. Platforms can use various available identifiers and signals to establish relationships, with the confidence and coverage of those relationships varying with the available data.

How is identity resolution different from cookie matching?

Cookie matching typically connects cookie-based identifiers between specific advertising systems. Identity resolution is broader: It can connect multiple types of identifiers across devices, browsers, sessions, and environments into a persistent record. Cookies can, therefore, be an input into identity resolution, rather than the complete identity solution.

What is a unified ID?

A unified ID is the single persistent record produced by identity resolution, representing one person across all the identifiers connected to them. It can also be called a cluster. Because the record can contain multiple identifiers, it can persist even as individual identifiers change or expire.