Identity resolution is the function that unifies the fragmented signals one customer leaves across devices, sessions, and channels into one persistent record. The output is an identity graph: each node is a person, each edge an identifier known to belong to them.
The identity graph
What an identity graph holds
Picture one shopper: she browses the mobile web (first-party cookie), logs into the iOS app (hashed email, IDFA), then opens a post-purchase email on desktop (same hashed email, fresh cookie). One node, four edges — any future session on those identifiers links back.
Inputs are hashed emails and phones, mobile device IDs, first-party cookies, account IDs, and probabilistic signals like fingerprint plus IP — the substrate is first-party data. Deterministic matches use exact identifier overlap; probabilistic matches infer from non-unique signals. Production graphs blend both and weight deterministic higher.
Where the function lives
Identity resolution is a job, not a product category. Several systems perform it:
- A CDP, as a core capability.
- A warehouse with reverse ETL, on stored event and order data.
- A clean room, in a privacy-bounded environment shared with a partner.
- An ad platform like Meta or Google, in its own graph for delivery and measurement.
A brand of meaningful scale usually runs more than one, with resolved identity from one feeding the next.
Why the function matters now
Apple’s ATT cut IDFA access on opted-out devices, browsers compressed third-party cookies, and ad platforms moved to modeled conversions. How well a brand resolves identity inside its own surfaces — and how that identity flows back via CAPI or a clean room — now decides whether attribution and audience targeting work.