Identity resolution

Recognizing that two different identifiers — an anonymous cookie, a later login — belong to the same real person.

Identity resolution is the process of recognizing that two different identifiers — an anonymous cookie set before a visitor logs in, and the known user ID they log in as — belong to the same real person. Get it wrong and most experimentation stacks quietly double-count that visitor: once as an anonymous assignment, once as a logged-in one.

The safe way to link identities is to record a relationship, not to rewrite history: an assignment already made under the anonymous key keeps pointing at that key forever, and a read-time lookup resolves a known key to its canonical identity and searches the whole linked cluster instead. See why anonymous-to-identified linking silently corrupts your experiment results for why re-keying is the wrong fix.