Glossary
Identity Resolution
Identity resolution is the process of connecting fragmented data signals from multiple sources, devices, sessions, and identifiers into a unified profile for a single individual or account. In marketing attribution, it is the prerequisite for every model: if the same buyer appears as a dozen separate anonymous IDs across different browsers, devices, and sessions, the attribution model is running on broken input data regardless of how sophisticated the credit distribution logic is.
The identity fragmentation problem
A B2B buyer's interaction with a vendor might span a laptop at work, a phone on the commute, a home computer in the evening, and a work computer at a conference. Each device, unless linked by a login event or other identifier, generates a separate anonymous ID in the analytics system. The same person generates multiple profiles, and their touchpoints are scattered across all of them.
Add to this the cross-channel fragmentation: an email click, a paid social ad click, and a direct website visit may all arrive from the same person but with different session cookies, different referrer headers, and different IP addresses. Without a resolution mechanism, each appears as a separate visitor.
Deterministic vs. probabilistic resolution
Identity resolution methods fall into two categories. Deterministic resolution uses a known identifier that is the same across all touchpoints, typically an email address, user ID, or hashed phone number. When a visitor submits a form, logs into a product, or clicks an email link, the email address links all prior and subsequent sessions to the same CRM contact with certainty.
Probabilistic resolution infers that two profiles belong to the same person based on shared signals: same IP address and device fingerprint, similar behavioral patterns, proximate time periods. Probabilistic matching is less certain than deterministic resolution but extends coverage to sessions where no identifying event has occurred. The combination of both approaches produces the most complete identity graphs.
Deterministic stitching
Email-based stitching is the gold standard for B2B identity resolution. When a CRM contact's email appears in an event, all sessions with matching first-party IDs are merged into one profile.
Probabilistic matching
IP address, device fingerprint, time patterns, and behavioral signatures are used to infer that anonymous sessions belong to the same person before they have identified with an email.
Account-level stitching
In B2B, individual contact profiles must be further grouped into account-level journeys so that a deal's full buying committee activity is visible as a single coherent unit, not separate individual records.
Privacy by design
A well-designed identity graph stores hashed identifiers rather than raw PII, limits cross-device matching to first-party consented signals, and retains data only for the period necessary for attribution.
The role of identity resolution in cookieless tracking
Third-party cookies served as a crude, shared identity layer across the web. When they stopped working reliably, the identity resolution problem became the central challenge in digital attribution. Without a shared cross-site identifier, each website must build its own first-party identity layer from scratch using the signals available within its own domain.
This is where cookieless tracking and identity resolution intersect. First-party session tokens maintain cross-session continuity within a single domain; server-side event capture preserves events that client-side blockers would suppress; and deterministic email matching provides the anchor that links anonymous sessions to known CRM records.
For B2B attribution, the completeness of the first-party data feeding the identity layer directly determines model accuracy. See first-party data for how to build the collection layer that feeds identity resolution.
Identity resolution in AttriByte
AttriByte builds its identity graph warehouse-native. Deterministic resolution stitches sessions to CRM contacts via email-based matching across form fills, email clicks, and product logins. Probabilistic matching fills gaps using IP, device signals, and behavioral patterns for sessions that precede identification. Multiple contacts at the same company are collapsed into a unified account journey before any attribution model runs, which is the correct unit of analysis for multi-touch attribution in a B2B context.
The identity graph is stored in your own warehouse instance. No customer identity data flows through AttriByte's infrastructure, which keeps the graph within your GDPR and CCPA compliance perimeter and makes it portable if you change attribution vendors.
Related glossary terms
Attribution starts with accurate identity
AttriByte's warehouse-native identity graph stitches anonymous sessions to known accounts before any attribution model runs, so your credit distribution reflects actual buyer journeys.