Clean up duplicate customer records before you automate anything — AI Use Case | Ava Digital
← All AI use cases
AI Use Case · Operations

Clean up duplicate customer records before you automate anything

A one-time sweep that flags likely duplicate customers in your CRM or job software and proposes which records to merge. You approve each merge; nothing is deleted without a click.

How it works

  1. Export your customer list to a Google Sheet with name, phone, email, address and last job date.
  2. Normalize phone numbers and emails with simple formulas, then group rows that share a phone, email or address.
  3. An AI step reviews each group and labels it as same person, same household or different people, with a one-line reason.
  4. You review the labels in the sheet and mark the ones to merge; merge them in your CRM yourself or with an import.
  5. Keep a short rule for new records, such as search by phone before creating a customer, so duplicates stop returning.

The guardrail

A person approves every merge and the AI never edits the CRM, since a wrong merge can mix up two customers’ history.

Tools that do this: Google Sheets, your CRM or field-service software export, an AI model (Claude or ChatGPT)

Prompt to start with
You are helping a small business clean up its customer list. Below is a group of records that share a phone number, email or address. Decide whether they are the same person, the same household, or different people.

Records:
[paste 2-5 rows with name, phone, email, address, last job date]

For each group return: LABEL (same person / same household / different people / unsure), CONFIDENCE (high / medium / low), a one-sentence reason using only the data shown, and which record to keep as the primary (most recent job date, most complete fields). If the data is not enough to tell, say 'unsure' rather than guessing. Do not suggest deleting any record; only recommend merges for a human to approve.
Get a free audit We’ll set this up for your business, end to end.