A vendor can tell you that its identity solution reaches millions of people. Then you hand over your own customer audience and discover that far fewer records actually match. That gap between reach claimed and audience matched is where many identity conversations get difficult.
Identity resolution advertising is the process of recognizing the same person, household or device across different environments so a buyer can use that connection for targeting, frequency management or measurement. The catch is that different identity approaches solve different parts of that problem and none should be treated as universal coverage.
The practical question is no longer which identity solution wins. It is how much of your defined audience each approach actually matches and what you have to pay and operate to get that coverage. That is the question this article focuses on: match rate against a defined universe rather than headline reach.
What identity resolution is trying to do
Identity resolution in advertising connects identifiers or signals from different places to determine whether they relate to the same person, household or device. That connection can support three different jobs, and buyers should treat them as separate problems.
| Job | What the buyer wants |
| Targeting | Find and reach an intended audience |
| Frequency control | Avoid counting or serving ads to the same person incorrectly across environments |
| Measurement | Connect exposure or audience records to outcomes |
The important part is that one identity approach may work well for one job and less well for another. A system that helps a buyer find an audience does not automatically provide the same level of visibility for measurement or frequency control. The useful question is therefore what the identity method actually supports in the media workflow being evaluated.
That distinction matters when comparing identity solutions. The technology can look similar on paper while the practical use case is different. For a media buyer, the job comes first: identify the audience, control exposure or measure the result.
The approaches and what each actually covers
Identity solutions use different signals and work in different environments. That changes how much of an audience a buyer can match and what the identity is actually useful for.
Authenticated or deterministic identifiers
These start with an identifier a user or customer has supplied or authenticated, such as an email address or phone number, where permitted. When the same identifier is legitimately available in two environments, it can provide a direct connection. Coverage still depends on authentication, consent and where that identifier is available.
Probabilistic or modeled identity
These approaches use multiple signals to estimate whether records relate to the same person or household. They can extend addressable coverage when direct identifiers are unavailable. The tradeoff is uncertainty because the connection is inferred and depends on the signals and methodology used.
Publisher and platform first-party identity
A publisher or platform can use its direct relationship with users for advertising or measurement within its own environment and approved partner workflows. That identity does not automatically become a universal ID across unrelated properties.
Contextual targeting
Contextual targeting does not resolve identity. It uses the content or environment to make an advertising decision when identifying an individual is unnecessary. It is therefore an alternative to identity-based targeting rather than another identity solution.
Clean-room matching
Clean rooms let two parties compare first-party data in a controlled environment for uses such as audience activation or measurement. The underlying records do not have to be openly exchanged. An identity graph can connect identifiers or signals into a broader identity structure, but its usefulness still depends on where those connections work.
Why match rate matters more than reach
A high reach number tells you how many people a solution says it can potentially address. Match rate tells you how much of a defined audience it actually connects with. That makes match rate more useful when you are deciding whether an identity solution can support a specific campaign.
What is a good match rate?
There is no universal good match rate. The number only means something when you know the source audience, destination, identifier, matching method and denominator used to calculate it. A simple formula is:
Match rate = matched records ÷ records in the defined source audience
The denominator can differ between systems. For example, current PAIR documentation defines match rate as the percentage of a user list’s membership that matches the corresponding publisher’s first-party data.
A simple example
Say a buyer has 1,000,000 consented customer records and a solution matches 40% of that defined audience with a selected publisher. The result is 400,000 matched records for that particular comparison.
That 40% does not mean the solution reaches 40% of the internet. It means 40% of the buyer’s defined audience matched in that specific environment.
So the vendor question should be simple: What is your match rate against my defined audience and what exactly is the denominator? Then ask for the matched count, eligible universe and methodology used to calculate it.
Clean rooms and what they are actually for
A clean room lets organizations work with data from different parties under controlled rules without simply exchanging their underlying datasets. In advertising, that can mean comparing an advertiser’s first-party audience with a publisher’s data for activation or measuring whether exposed users later took an action.
Audience activation
For activation, a clean room can help an advertiser and publisher identify an overlapping audience without either side handing over its underlying customer records. The matched result can then support an approved advertising workflow.
Measurement
Measurement is another major use. An advertiser and publisher can compare exposure and conversion records in a controlled environment to understand campaign outcomes. Clean rooms are therefore not measurement-only, but measurement is one of the important jobs they can perform.
The operational cost
The technology also creates work around it. Teams may need to prepare and format data, manage permissions, define which queries are allowed and reconcile results across partners. Different partners can also require separate implementations, which adds operational work even when the underlying goal is similar.
The exact setup varies by clean room, so buyers should ask what data can enter the environment, which queries are permitted and what comes out. A clean room does not automatically produce a higher match rate or remove the need to meet applicable privacy requirements.
How many identity solutions does a buyer end up supporting?
There is no fixed number of identity solutions a buyer needs. The practical answer depends on where the audience exists, which identifiers are available and whether the job is targeting, frequency management or measurement. A buyer working across web, apps and CTV may therefore need different identity routes for different parts of the media plan.
Why the stack gets fragmented
Publishers and platforms can use different identity environments. Some audiences are authenticated while others are not, and the signals available in CTV can differ from those available on the web. Measurement can also require a different matching workflow from audience activation.
That means supporting multiple routes can become necessary when one approach does not cover the environments a buyer actually uses. IAB Tech Lab’s guidance describes identity solutions across web, mobile apps and CTV while also documenting ID-less approaches for environments where identifying signals are unavailable.
Where the cost shows up
The expense is not only the platform fee. Teams can also take on data onboarding, integrations, clean-room work, identity reconciliation and different reporting requirements. Audience definitions can also get duplicated when separate systems describe the same customer base in different ways.
That creates a reporting problem: two systems can both report reach while using different definitions of the addressable audience. Platform support therefore becomes part of the buying decision, which is why the DSP selection question matters.
Where privacy regulation constrains all of it
Identity resolution does not have one privacy rule that applies everywhere. The legal basis for processing, consent requirements and limits on data sharing depend on the jurisdiction, technology and processing involved. The same identity workflow can therefore face different requirements in different markets.
Consent and lawful basis
Buyers need to understand what data is being used, why it is being processed and which parties receive or access it. Depending on the technology and jurisdiction, that can involve consent or another lawful basis under the applicable privacy framework.
For online advertising, current UK guidance says storage and access technologies used for advertising require consent in the situations covered by its rules. It also addresses tracking, profiling and measurement separately.
Why the market matters
The same identity approach can involve different obligations when it crosses jurisdictions. Data sharing, user rights, retention and permitted purposes can all affect how an identity solution is implemented.
That makes privacy a practical part of identity evaluation rather than a final compliance check. Buyers should establish the applicable requirements for each market before matching or activating data.
Treat this section as jurisdiction-specific and time-sensitive. Privacy rules and regulatory guidance can change, so a current legal review is needed for any real implementation.
What a buyer should actually do
The safest way to evaluate an identity solution is to ignore the headline reach and audit the match itself. Start by defining the exact universe you need to match, whether that is your CRM audience, logged-in users, purchasers, subscribers, site visitors, or an addressable media audience.
Then ask for the denominator. If a provider says it can reach 200 million people, ask, “200 million out of what eligible universe?” Without that denominator, the number tells you very little.
Next, require the source count, matched count, match rate, matching method, destination, and time period. These details show what the claimed reach actually represents. Then calculate your cost per matched user:
Cost per matched user = total identity cost ÷ matched users
Keep media or platform costs separate if they are included in the calculation. Finally, test the identity route against the specific job you need it to solve before integrating everything. Compare the results using the same audience definition and measurement period.
| Question | What to demand |
| What is the audience? | Exact source universe |
| What is the denominator? | Eligible records |
| How many matched? | Matched count |
| How was it matched? | Deterministic, probabilistic, or other |
| What does it cost? | Total cost and cost per matched user |
Final Verdict
Identity is unlikely to become one universal solution unless the market develops much broader interoperability. For buyers, the more useful reality is that different identity approaches solve different jobs.
The documented use cases are already specific. These include first-party matching, clean-room activation, clean-room measurement, and platform-specific identifiers. What remains less certain is how broadly any one approach will work across the wider ecosystem, whether identity solutions will consolidate, and what future regulation or platform policies will allow.
That makes the headline size of an identity solution less useful than the quality of its match. A buyer needs to know which audience was measured, what the eligible universe was, how many users matched, and how the match was produced. The question to end with is simple: What is your match rate against my defined audience?
