// CRM data quality
Keep CRM records useful for the next conversation.
An illustrative process for finding duplicates, incomplete records, and deals without a clear next action.
Automation Example. Example workflow, not a customer case study.
The starting situation
Contacts arrive from forms, imports, and manual entry. Names vary, records overlap, and ownership becomes unclear. This workflow creates a reviewable cleanup queue without silently rewriting the team’s history.
The proposed workflow
- SystemRead a limited set of records and check agreed required fields.
- AutomatedNormalize permitted formats while preserving original values.
- AISuggest possible duplicates with the fields supporting each match.
- AutomatedFlag stale deals and missing next steps for the assigned owner.
- Requires approvalApply approved changes, log the outcome, and leave unresolved items in the queue.
Where a person decides
A person reviews ambiguous matches and approves merges or destructive changes. Matching a company name alone is insufficient. Conflicting owners or consent records are flagged; recent edits must be checked before applying an update.
Tools and data
The existing CRM, its available API or export, and an agreed field dictionary. The pilot needs clear source-of-truth rules and a recoverable record of changes.
What a pilot should measure
Track confirmed duplicates, completeness of required fields, incorrect match suggestions, and deals without a next step. Review sampled changes to ensure apparent cleanliness does not hide lost information.
What to bring to an audit
Bring a redacted sample of records, definitions of required fields, and examples of true and false duplicates. Start with one list and a review-only run before allowing writes.
Show us the workflow your team still handles manually.