Playbooks · 4 min read

Dedupe before delivery: keeping the feed clean

Dirty data is not a cleaning problem, it is a delivery problem. Records should be deduped before they reach the CRM, not after.

Match on stable fields

We match on postcode and director names - the fields that survive rebrands and trading-name changes. No fuzzy company-name guessing.

If the company already exists in your CRM, the feed skips it. Merging downstream is where duplicate hell starts.

Every skipped record is logged with the reason and the matched CRM id, so "why didn’t this come through" has an answer.

Try it this morning

Run a week’s feed in skip-logging mode and review what would have been duplicates.