How to anonymize an Excel file in three minutes
- Drop your file. Your browser reads it in your computer's memory. No copy is made anywhere else, and your original file stays intact.
- Pick a treatment per column. The tool spots columns that look like names, emails, phones, addresses, dates of birth, postal codes or social insurance numbers, and suggests a treatment. Three quick presets cover the common cases; you have the final say.
- Review and download. The preview shows the first rows before and after. Warnings point out what could still allow someone to be re-identified. You get a new Excel or CSV file and, if you want, a report describing the treatments.
When to use it
- Before sending an export to a consultantA CRM, payroll or accounting export holds more personal information than needed to analyze sales or automate a report.
- Before pasting a table into ChatGPT, Copilot or ClaudePseudonymize identity columns and scrub the notes: the assistant can still count, group and compare.
- For a vendor or subcontractorShare the columns needed for a test or an integration, without the identities.
- For a demo or a training sessionShow a real file, with its structure and amounts, without exposing your customers or employees.
Anonymized, de-identified, pseudonymized: what Quebec's Law 25 says
“Anonymize” is used everywhere, but in Quebec it has a precise meaning. Since Law 25, the Act respecting the protection of personal information in the private sector distinguishes two levels:
- De-identified: the information no longer allows the person to be directly identified. That is what pseudonymization produces. The file remains personal information, subject to the Act.
- Anonymized: it is reasonable to expect that the information irreversibly no longer allows the person to be identified, directly or indirectly. This level is governed by a regulation and requires a re-identification risk assessment.
In practice, this tool helps you de-identify a file and apply data minimization. To aim for anonymization in the legal sense, also remove or generalize the columns that, combined, can identify someone: full postal code, date of birth, sex, job title. The warnings panel flags them. The same logic applies under PIPEDA and the GDPR. This page is not legal advice.
Why a plain hash is not enough
Many tools replace a name with its SHA-256 digest. The problem: anyone can hash a list of common names and find the matches. This tool uses an HMAC with a key only you know. Without the key, identifiers can neither be recomputed nor reversed.
Verify for yourself that nothing is sent
- Disconnect from the Internet (Wi-Fi or cable) after opening the page, then use the tool: everything works.
- Open the developer tools (F12, Network tab): no request leaves during processing. The counter at the top of the page performs the same check. The only exception is the Excel reading library, loaded from this same site when you drop an Excel file.
- Read the page's security header: its
connect-src 'none'policy forbids the browser from making any outbound connection from this page, for us or anyone else.
Frequently asked questions
Is it really free?
Yes, with no account and no usage limit. The tool is offered by D3 Conseil, a Montreal firm specializing in process automation and data. If you need more (documented anonymization, unstructured data, large volumes), that is a project we take on.
Which formats and file sizes?
Excel (.xlsx, .xls) and CSV, comma- or semicolon-separated, in UTF-8 or Excel's Windows encoding. Since everything runs in your browser, the limit depends on your computer: up to about 250,000 rows is comfortable.
Are formulas, other sheets and formatting kept?
No. The exported file contains the values of the selected sheet, without formulas, colours or other sheets. This is deliberate: a formula, a hidden sheet or a pivot-table cache can hold data you thought you had removed. For a multi-sheet workbook, process each sheet in turn.
Can the original data be recovered?
Not from the transformed file. Identifiers are computed with HMAC-SHA256 and your key: they cannot be “decrypted”. Only the mapping table, if you choose to download it, links them back. Keep it on your side.
Do I need to send the key with the file?
No. The key only serves to get the same identifiers from one file to the next. Write it down if you prepare several files that must be linkable; otherwise you can forget it.
Why not just delete the columns in Excel?
That is perfect when the column is useless. But if you need to know that two rows concern the same customer, or to link two files, replace it with a stable identifier rather than deleting it. And Excel sometimes keeps hidden copies: hidden sheets, history, pivot-table caches.
What is the de-identification report for?
It lists the columns, the treatment applied to each, the methods used and the warnings raised. It is the document to hand to your privacy officer or to keep with the shared file, to show what was done.
What next?
If this file is an export someone reworks by hand every week or month, there are probably hours to be saved. D3 Conseil automates this kind of processing for Quebec SMBs: Excel reports, Sage exports, Power BI dashboards, migration to Odoo, and Law 25 compliance of data flows.