Contacts, leads, or deals. Finds missing fields, duplicate records, and enum drift — scores and ranks what to fix first.
Everything runs in your browser. No file is uploaded.
—·—·—
No data transmitted
Deduplication configuration
Switch between configurations to see how field mapping and threshold choices change what the audit finds. All three run on the same data.
Standard
Algorithm defaults, no field context
Email-only blocking at threshold 0.95. Only catches exact email collisions — records sharing a name and company under different email addresses are invisible to this run.
Works for obvious duplicates. Misses the structural ones that actually cause decisions to go wrong.
Configured
Field roles mapped, fuzzy matching on
Email marked as identity field. Name and company used as co-signals within email-domain blocks. Jaro-Winkler at 0.90 — catches near-matches Standard cannot see.
Broader recall. Score drops not because the data got worse, but because the audit is looking harder.
Custom · active
Tuned to this dataset's structure
Company-aware blocking. Owner and company weighted as co-signals alongside email. Threshold at 0.87 — calibrated to how contacts in this CRM are actually named and structured.
Lowest score of the three. Most actionable output. The score reflects what was already broken.
Data health
Dimensions
Completeness—
Uniqueness—
Consistency—
Freshness—
no date field marked decision-critical
Prioritised fixes
This tool surfaces and ranks. It does not fix — the decisions stay human.