Data Cloud Identity Resolution: Matching and Reconciliation Explained

Identity resolution in Data Cloud is two decisions wearing one name. Match rules decide which source records describe the same person. Reconciliation rules decide, once they are joined, which value wins. Teams argue about the first and are surprised by the second.

Match rules

A match rule is a set of criteria — exact or fuzzy — evaluated across source profiles. Loose rules over-merge: two people in a household become one, and someone receives a message about an order they did not place. Tight rules under-merge: the same person appears three times, and your counts are wrong in the other direction.

Over-merging is the worse failure. It is visible to customers, it is a privacy problem, and it is hard to unpick after activation.

Reconciliation rules

Once records are unified, something must choose the surviving email address, name, and consent flag. Most frequent, most recent, or source priority — each is defensible, and each gives a different answer. Consent in particular should almost never use "most recent wins" without thinking it through.

Testing before you trust

• Assemble a deliberate test set: obvious duplicates, household members, a person who changed email, a person who changed name, and a shared device.
• Run the ruleset and check the unified profile for each by hand.
• Count unified individuals before and after a rule change — a sudden drop is an over-merge, not a win.
• Re-check consent fields specifically, separately from everything else.

Conceptual explanation based on current Data Cloud documentation. Confirm the exact rule types and limits available in your org before designing around them.