Rewrites values in your file
Clean & Anonymize
The twenty things wrong with every export, found and fixed in one previewable pass — and the personal data in it, found and dealt with in the same pass. Every finding is a deterministic rule over the column, explainable in one sentence. There is no model and no API key.
Windows 10/11 (x64) · Free · No account
A modified copy is written to DataTray's output folder, plus a rejects file for anything that would not convert. The original is never edited in place.
What you can do
What people use it for — Clean & Anonymize
The jobs this one actually gets opened for.
- Cleaning a CRM export before it goes into anything that matters.
- Splitting one column into several — the text-to-columns job, on a file Excel cannot open.
- Masking or hashing customer emails and phone numbers before sharing a file with an agency.
- Fixing dates that arrived in four different formats in the same column.
- Sending someone a representative sample of a file without sending them the file.
What it does
Everything Clean & Anonymize handles
Everything this tool handles, and how it handles it.
It finds the problems first
Numbers stored as text, mixed date formats, stray whitespace, inconsistent casing, twelve spellings of “California”, mojibake, mixed thousands separators, currency symbols in a numeric column, empty and constant columns, ragged rows, and leading zeros that have already been lost.
An operation stack you can reorder
Trim, collapse whitespace, normalize case, coerce types, normalize dates to ISO-8601, fix locale-aware number separators, find and replace, split or merge columns, rename and drop, fill blanks. Applied in order, previewable, and reversible before anything is written.
Personal data, found deterministically
Email, phone, national ID, payment card (Luhn-checked, so a 16-digit order number is not flagged), IBAN (mod-97), IP address, date of birth, address shape and person names. Every finding comes with a confidence and a matched-row count, at column level.
Redact, mask, hash, tokenize or generalize
Per column. Hashing shows and saves the salt, because an unrecorded salt makes the output unjoinable and nobody notices until later. Tokenizing gives a stable pseudonym per value, so joins still work afterwards.
Offline enrichment, from bundled tables
Split a full name into parts, normalize country and state names both directions, check postcode shape by country, put phone numbers into E.164, validate email syntax. Small versioned lookup tables in the installer — data, not intelligence, and the app never claims otherwise.
Save it as a recipe
Someone who just fixed twelve things on this month's export will need to fix the same twelve on next month's. Save the stack and re-run it, or hand it to a watched folder.
How it works
From your file to the result
- 1
Open the file
DataTray scans it and lists what it found, severity first, with a count per finding and a preview of the affected rows.
- 2
Build the stack
Click the fixes you want. Personal-data findings are grouped and visually distinct, so anonymization is right there rather than buried. Reorder anything; preview as you go.
- 3
Run it, and keep it
Out comes the cleaned copy, a rejects file, and a report that states what was changed and — for a compliance reader — what was deliberately left untouched.
Under the hood
Formats, limits and defaults
Everything here is a deterministic rule over column statistics, computed locally in milliseconds.
- Detection
- Regex plus checksum validation — Luhn for cards, mod-97 for IBAN — to keep false positives usable
- Anonymization
- Redact · partial mask · SHA-256 hash with a recorded salt · stable tokenize · generalize · drop
- Sampling
- n rows or n%, optionally stratified by a column, and optionally anonymized in the same pass
- Outputs
- Cleaned copy in any supported format · a rejects file · the Clean & Anonymize Report
- AI
- None. No model ships in the installer and there is no key to enter in Settings
Nothing is ever decided for you. Every personal-data finding is presented for confirmation before a single value is rewritten — a tool that silently decides what counts as personal is worse than no tool.
The rest of the belt
The other four tools
Questions
Clean & Anonymize, answered
Get it free — Clean & Anonymize
Clean & Anonymize ships in DataTray, free for Windows 10 and 11. No account, and nothing uploads.
Windows 10/11 (x64) · Free · No account