Find near-duplicate customers with fuzzy SQL matching

Marketing swears the customer list is clean, and 'Jon Smith' at one address sits three rows away from 'John Smith' at almost the same one.

AdvancedRead 8 minDoing: an hour or two for a first pass on a small table, faster once you reuse the same blocking keys and weights on the next one8 steps

▶ Start

What you'll have at the end

A scored list of likely duplicate pairs, above a chosen similarity threshold, ready for someone to confirm and merge.

Before you start

You need

A customer table sitting in a real database, with at least a name, a postal code, and a phone number or email address on every row, plus permission to create new tables from it.

Not covered

Keeping a written log of the cleaning decisions this run makes, and telling apart two real, different people who happen to share a name and an address, such as family members at one house.

Leans on

8 steps

Where to go next

Write a reproducible data cleaning log

when you also want a written trail of every cleaning decision made along the way, which this recipe deliberately leaves out.

Checked 26 Aug 2026

Part of the Data Cleaning and Validation for Analysts cookbook