Finding data mismatches before fixing the wrong record

Start with the source of the mismatch

A contact's role, status or license membership could disagree between systems. A support request showed the symptom, but the cause could be an association, an identity mismatch or an earlier step in another system.

How I designed the comparisons

Two datasets compare contacts in opposite directions: learning platform to HubSpot, and HubSpot to learning platform. Each preserves its starting population so records missing on the other side remain visible.

The third compares licenses and member counts. I accounted for multiple source rows per license to keep distinct license counts separate from row counts.

I defined alignment logic in the datasets and used report filters to select findings. The dashboard brings missing contacts, missing associations and inconsistent statuses together across the population.

One view of the exceptions

Teams can investigate an exception and its source before changing the record named in a support request. The dashboard detects differences; it does not automatically repair records.

Before correcting a record

A finding starts an investigation. Check identity, associations and the authoritative source before proposing a repair or changing access.

An unmatched email can reflect a changed identity, a missing record or a delayed refresh. For snapshot-sensitive findings, wait for the next Snowflake refresh and check whether the difference persists.

License member-count findings still need validation. Clearer source freshness is also open work, so the dashboard cannot guarantee realtime alignment. Semi-automated and fully automated repair remain follow-up work.