Deduplicating a lead list across formatting variants

The challenge

The same person enters a lead database more than once under slightly different names, emails, or companies. Duplicates split activity history, double-count pipeline, and cause duplicate outreach.

Data before & after

[
  { "name": "Bob Smith", "email": "bob@acme.com" },
  { "name": "Robert Smith", "email": "bob@acme.com" },
  { "name": "Bob Smith", "email": "b.smith@acme.com" }
]

[
  {
    "canonical": { "name": "Bob Smith", "email": "bob@acme.com" },
    "duplicates": 2,
    "confidence": 0.94
  }
]

The workflow

1. 1Normalize records first so trivial formatting differences do not hide matches.
2. 2Run the set through Dedupe to group likely-duplicate records with a confidence score.
3. 3Auto-merge high-confidence groups; route ambiguous groups to review.

Capabilities used

Implementation

Normalize with Repair, then send the batch to Dedupe. Use the confidence score to decide auto-merge vs. review; use Match to compare a single pair when you need a targeted decision.

curl -X POST https://zapinner.com/api/v1/dedupe \
  -H "Authorization: Bearer $ZAPINNER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "records": [ /* your lead rows */ ] }'

Technical architecture

• Normalize-before-dedupe removes formatting noise so matching keys on real differences.
• Confidence thresholds let you separate safe auto-merges from human review.
• Match handles the single-pair decision inside a larger workflow.