The pattern that resolves pre-identified behavior to a person once they identify themselves through login, form submission, or purchase. Connects session-level behavior captured under an anonymous identifier (typically a device ID or session cookie) to a stable user record, carrying the prior behavioral context forward. Foundation of nearly every personalization, attribution, and conversion analytics recipe.
The pattern that connects the same person across phone, laptop, tablet, and other surfaces. Combines deterministic signals (shared logins, hashed email matches) with probabilistic signals (device graphs, behavioral fingerprints) depending on consent state and data availability. Required for any recipe needing continuity of context across the customer's actual device usage rather than treating each device as a separate user.
The pattern that resolves individual identifiers up to a shared buying unit: a household, family, B2B account, or shared subscription. Necessary for verticals where the buying decision isn't individual: family streaming services, B2B procurement with buying committees, financial products with joint holders. Distinct from individual identity resolution because the unit of value isn't the person.
The pattern that handles what happens when identifiers age out, consent is withdrawn, cookies expire, or customers go dormant. Defines the graceful degradation path: which signals remain usable, which fall back to anonymous, which are deleted entirely. Most martech stacks treat identity as permanent and break in interesting ways when it isn't.
The pattern that defines how new edges land in the identity graph: synchronous at the point of resolution, deferred in a downstream job, or some hybrid that prioritizes load-bearing edges. The write semantics determine whether the rest of the stack can trust the graph at decision time, and most stacks default to deferred-batch in a way that fails the use cases that motivated the graph in the first place.