"Just remove the name column" is rarely enough. Real anonymization means thinking about what combination of remaining columns could still identify someone, not just the obviously personal ones. See also our broader guide on CSV security and privacy.

Three Techniques, and When Each Fits

Removal

Drop the column entirely. Use when the recipient has no legitimate need for that data at all, the safest option, but only works if the column truly isn't needed.

Masking

Show only part of a value, e.g. an account number as ****1234. Use when the recipient needs to recognize or reference the record but doesn't need the full value.

Pseudonymization

Replace a real identifier with a consistent fake one, the same person always gets the same substitute ID. Use when relationships across rows matter (e.g. tracking one customer's repeat purchases) but the real identity doesn't.

The Mistake: Only Removing the "Obvious" Columns

Dropping name, email, and phone feels like enough, but a combination of seemingly harmless columns, zip code, birth date, and gender, has been shown to uniquely identify a large majority of the US population on its own. This is called a "quasi-identifier" problem: no single remaining column identifies someone, but the combination does.

Before sharing: ask not just "does any column directly identify someone" but "could someone with outside knowledge use a combination of the remaining columns to figure out who a row is about."

Generalization: Reducing Precision Instead of Removing

Sometimes the fix isn't removing a column, it's making it less precise. An exact birth date narrows a person down a lot; a birth year alone narrows much less. A precise home address is identifying; a zip code or city usually isn't on its own. Generalizing quasi-identifying columns (reducing precision) preserves analytical usefulness while reducing re-identification risk.

A Practical Pre-Share Checklist

Every direct identifier (name, email, phone, ID number) is removed, masked, or pseudonymized
Quasi-identifying combinations (date + location + demographic details) have been considered together, not just column by column
Precision on sensitive fields has been reduced where full precision isn't actually needed for the recipient's purpose
You know exactly who receives the file and can account for where copies will end up

Doing This in How To CSV

The Anonymizer tool handles masking and pseudonymization of sensitive columns, entirely in your browser, before you export or share a file.

Preparing a file to share externally?

Anonymize sensitive columns without uploading the original file anywhere.

Anonymize Your Data

Turn this into a saved workflow

Create a free account to save the steps from this guide as a reusable workflow and re-run it on any file, from any device.

How to Anonymize CSV Data Before Sharing It