---
name: crm-duplicate-record-review
description: Finds and explains likely duplicate leads, contacts, accounts, and opportunities across CRM exports without merging or deleting records. Use for CRM deduplication, duplicate review, data hygiene, account matching, or pre-migration record cleanup.
license: Apache-2.0
metadata:
  adlass.categories: "sales-crm/pipeline, sales-crm/lead-research"
  adlass.industries: ""
  adlass.tags: "crm-cleanup, duplicates, record-matching, data-hygiene, accounts, contacts"
  adlass.adaptation: "mapping"
  adlass.source: "n8n:1822, n8n:664, n8n:1792"
  adlass.version: "1"
---

# CRM duplicate record review

## Purpose

Identify probable duplicate CRM records and provide a safe, evidence-based disposition for review. Preserve the source records and show exactly why two records were considered related.

## Scope

Accounts, contacts, leads, opportunities, identity fields, ownership, activity, stage, timestamps, and conflicting values.

**Excluded:** merging, deleting, overwriting CRM records, or resolving ambiguous identities by guesswork.

## Data basis

- CRM exports or record tables in scope.
- Company matching rules and field mapping.
- Optional reference table for known account aliases.

## Result

A duplicate-candidate sheet and a concise disposition report.

## Quality criteria

- Each candidate pair or group lists matching fields and confidence.
- Exact matches are separated from fuzzy candidates.
- Conflicting owners, stages, or activities are visible.
- Every input record receives a reviewed status.
- No source value is lost.

## Instructions

Apply exact identifiers first, then conservative combinations of normalized name, domain, email, phone, address, and account identifiers. Never use a shared company domain alone to merge people. Treat different opportunities at one account as separate unless the opportunity identity matches. Keep threshold values in the mapping rules.

## Adapt before use
Keep a clear audit trail throughout the run. Use the source wording and field values that support each material conclusion, and retain uncertainty when the evidence does not decide the issue. Separate a missing record from a negative result and a contradictory record from an exception. Prefer a short, prioritised result over unsupported completeness. Check that counts, identifiers, dates, and labels agree across the written result and the structured output. When a rule, source, or mapping is unavailable, name the limitation and explain how it affects interpretation. Do not silently infer ownership, approval, timing, or business intent.

- Map record types and identity fields.
- Define exact and fuzzy match thresholds.
- Add known aliases and the review-status vocabulary.