A visitor browses anonymously, later signs up in an app and contacts support from a different email address. Marketing may count three people; service may see only one. Identity resolution helps, but a mistaken merge can be worse than a missing link.
Two adults use one family email for shopping, while each has a separate loyalty account. A naive identity rule combines their purchases and preferences. That can lead to irrelevant messages and an inaccurate customer history. Conservative matching and an easy correction path are part of a responsible customer view.
Establish what can be linked
Separate authenticated IDs from inferred matches. Document how records may be joined and how consent limits use. Keep the source and confidence of each match. Shared devices and family phone numbers create traps for simple rules.
Identity decisions can affect customer service as well as measurement. If two people are wrongly merged, an agent could refer to another person's order. Build permissions and review around that risk. Keep a clear audit of how identifiers were linked and separated. A system should be able to undo a bad merge without losing the history of either person.
Be modest about attribution
Even with improved identity, teams rarely observe every influence on a purchase. Offline conversations and private sharing leave gaps. Use attribution as a decision aid, not proof of exactly why someone bought.
Audit both kinds of error
Review false joins and missed joins, especially around important journeys. Let customers and support correct account information. Report the share of activity that cannot be linked. Credible measurement makes its blind spots visible.
