Identity Resolution Audit (QA)

Identity Resolution Audit (QA)

1. 01

   Pull a 30-day sample of raw events and count distinct identifiers per known customer

   If the average profile carries more than three unmerged identifiers (email, device ID, loyalty number, hashed phone), the client's match logic is underperforming and every downstream segment inherits the error.

2. 02

   Map every source system feeding the CDP and label its primary key

   A composable setup such as Hightouch reads directly from the warehouse, so the warehouse primary key governs; a full-suite platform like Braze may ingest its own anonymous ID first and reconcile later.

3. 03

   Test deterministic matches before probabilistic ones

   Confirm exact email and authenticated login matches resolve cleanly, then measure how many additional profiles fuzzy matching adds and whether those additions survive a manual spot check of 50 records.

4. 04

   Document the merge and unmerge policy in writing for the client

   State who can force a merge, what happens to consent flags when two profiles combine, and how long an unmerge request takes. Agencies that skip this step absorb the blame when a client sees one person receive two offers.

5. 05

   Verify consent and suppression state travels with the merged profile

   A profile merged from a subscribed email record and an unsubscribed SMS record must not silently re-enable the suppressed channel. Check this against the client's regional rules before any activation goes live.

6. 06

   Run a duplicate-rate baseline and set a monthly threshold

   Record the current duplicate percentage, then agree a ceiling with the client (commonly under 2% for B2C retail). Anything above the ceiling triggers a re-audit rather than a quiet patch.

7. 07

   Log the audit result in the client's delivery file with the next review date

   Identity drift is continuous, so a one-time clean-up has no retainer value. Tie the next audit to a calendar date and attach it to the monthly reporting pack.