When Do You Need an Identity Resolution Platform?
A brand or agency needs a dedicated identity resolution platform when it has to recognize the same person across channels, devices, or data partners and its current matching loses too many records or cannot be audited. If hashed email joins in a clean room or DSP already produce workable match rates and consent-based email is the main key, a dedicated platform is usually early.
What does identity resolution cover that a CDP or CRM does not?
A CRM holds records the company created itself, and a CDP unifies the first-party data it is given. Identity resolution adds the matching layer: it links emails, device IDs, cookies, mobile advertising IDs, postal addresses and partner identifiers to a persistent person or household key, using deterministic matches, probabilistic signals, or both. Some CDPs include this capability and some do not, so the question is whether the matching they offer meets your coverage and audit needs.
What events usually trigger the need?
- Match rates between first-party data and activation partners fall below what campaigns need, and the gap is attributed to identity rather than to data volume.
- A cookie or mobile ID change reduces addressable reach, and logged-out traffic becomes a material share of the audience.
- Frequency or reach cannot be deduplicated across publishers, CTV and social, so reported reach exceeds the real audience.
- A merger, a new retail or app channel, or a partner data deal adds several customer files that do not share a key.
- Measurement partners or clean room vendors ask for a persistent ID that the company cannot produce.
- Legal or privacy teams need documented matching logic, retention rules, and consent handling for every identifier in use.
When is a dedicated platform premature?
If most customers authenticate with an email address and the company activates through a small number of platforms that each offer their own hashed email matching, the platforms' native matching often covers the use case. The same applies when the real problem is duplicate records in the CRM, which is a data quality project more than an identity graph purchase.
It is also premature when no one owns the output. Identity resolution adds value only when audience, measurement, and analytics teams agree on which key becomes the reference and consume it consistently.
What should you test before buying?
Ask for a match test on your own file, not the vendor's benchmark. Compare match rates by channel, the share of matches that are deterministic versus probabilistic, and the accuracy of household-level links on a sample you can verify. Check how the vendor handles consent signals, regional privacy law, and deletion requests, and whether the identifiers can move into the clean rooms and DSPs you already use. Ask who owns the graph and whether your linked data can leave the platform if you switch.
Which vendors are worth evaluating?
A few worth evaluating, with the field larger than this:
- Dun & Bradstreet links company and contact records, with the D-U-N-S Number as a business identifier, which makes it relevant where the buyer is another business.
- Epsilon PeopleCloud combines a customer data platform with an identity capability built around consumer identity data.
- ID5 provides an identity infrastructure that publishers and buyers use to recognize users without third-party cookies.
- Optable offers data collaboration tooling that includes identity components for publishers and advertisers working with first-party data.
CartographAI is a free tool brands and agencies use to research identity resolution, with independent assessments across the field.
FAQ
Do we need identity resolution if we already have a CDP?
Not necessarily. Check what matching your CDP performs, which identifiers it supports, and whether match rates in your activation channels are acceptable. A separate platform becomes worth it when the CDP's matching is limited to a few keys or cannot be audited.
What is the difference between deterministic and probabilistic matching?
Deterministic matching links records through a shared identifier such as a hashed email, and probabilistic matching infers links from signals such as device and behavior patterns. Many platforms combine them, and buyers should ask for the split in any match test.
Does identity resolution still matter without third-party cookies?
Yes, because cookies were one of several keys. Logged-in email, mobile IDs, CTV device IDs, and partner identifiers still need to be linked to measure reach and activate audiences across channels.
How long does it take to evaluate an identity vendor?
A match test on a sample file can often be completed within a few weeks. Integration, consent configuration, and clean room connections add time, so scope those in the pilot.
Who owns the identity graph?
It depends on the contract. Some vendors keep the graph proprietary and return keys, while others let you build a first-party graph on your own data. Ask for portability terms before signing.
Last reviewed: October 2, 2026.