Glossary · Business Analytics

Identity Resolution

also: identity stitching · identity graph · probabilistic identity · deterministic identity

Definition

Identity resolution is the process of linking pseudonymous user signals (cookies, device IDs, IP/UA fingerprints) into a coherent person-level view. Logged-in deterministic matches are typically 8% to 18% of traffic; probabilistic matches fill the remainder with calibrated confidence scores rather than absolute IDs.

Identity resolution sits at the center of the post-cookie measurement stack. The spectrum runs from deterministic (email hash, logged-in ID, with effective 100% confidence when present) through probabilistic (IP plus user-agent plus behavior, with confidence scores typically in the 0.55 to 0.85 range) to aggregated cohort approaches (Topics API, FLEDGE/Protected Audience). The 'match rate' problem is the dominant operational constraint: most consumer sites have 8% to 18% logged-in traffic, so probabilistic methods carry the bulk of stitching work. Data clean rooms (LiveRamp, Snowflake Clean Room, AWS Clean Rooms) let two parties match on hashed identifiers without exposing the underlying PII.

Essays on this concept