A call note before it goes in a shared channel
This note from a sales call contains an email, a mobile number, a postcode, an NI number entered by mistake and a bank account. Here Keep email domains and Keep the first half of postcodes are both ticked, so the shared copy still shows which business and which district it is about.
Every kind was found once. The person's name, Mel Hart, is not touched: names are not something a pattern can recognise reliably, so read the result and remove names by hand where needed.
Spoke to Mel Hart ([email protected], 07700 900456) about the stand at BL3 2AB. Her NI number QQ 12 34 56 C was on the form by mistake. Deposit from GB82 WEST 1234 5698 7654 32.
Spoke to Mel Hart ([EMAIL]@copperleaf.example, [PHONE]) about the stand at BL3 [POSTCODE]. Her NI number [NI NUMBER] was on the form by mistake. Deposit from [IBAN].
Redacting
Add the text
Paste into the box or press Open a text or CSV file for .txt, .csv, .tsv, .json, .md or .log files. The download keeps your file name with -redacted added.
Choose what to find
Six kinds can be switched on or off: email, UK phone, UK postcode, IBAN, NI number and payment card. A live count beside each option shows how many it matched.
Choose the replacement
[EMAIL] style labels say what was removed. [REDACTED] hides the kind as well. ████ blocks keep the original length.
Keep a little context if useful
Keep email domains leaves the part after the @. Keep the first half of postcodes leaves the outward code, such as BL3.
Read it, then share
Check the output for names, street addresses and anything unusual, then copy or download it.
How each kind is recognised
| Kind | Recognised by | Not caught |
|---|---|---|
| The usual name@domain pattern | Addresses written as name at domain dot com | |
| UK phone | A UK number that converts to +44 | Foreign numbers, numbers with words in them |
| Postcode | A valid UK postcode shape | Postcodes split across lines |
| IBAN | Country code, check digits and a passing checksum | Sort code and account number written separately |
| NI number | Two letters, six digits, a final A to D | Deliberately broad, so lookalikes are hidden too |
| Card number | 13 to 19 digits that pass the Luhn check | Partial numbers such as the last four digits |
When redaction helps
Sharing a sample of a lead list with a freelancer, pasting a support conversation into an AI assistant, posting a bug report with a real export attached, or sending a supplier an example of a file format: each is a moment when personal data can leak for no reason. Redacting first means the other person sees the structure without the details.
UK GDPR's data minimisation principle asks you to use only the personal data you need. Redaction is one practical way to do that, but it is not anonymisation: a redacted note can still identify someone through context. Treat the output as reduced, not as safe to publish.
Questions people ask
How do I redact personal information from a document?
Copy the text into the box, or open a text or CSV file, and tick the kinds of data to hide. Each match is replaced as you type. Read the result, remove any names or addresses the patterns missed, then copy or download the redacted copy. For PDFs, use a PDF redaction tool instead.
Does it remove names?
No. Names have no fixed pattern, so a tool that tried to find them would miss many and hide ordinary words. The redactor handles data with a recognisable shape, such as email addresses and NI numbers. Search for names yourself and replace them, for example with Person A, before sharing.
Will it work on a CSV export?
Yes. Open the CSV and every cell is redacted in place, so the columns stay where they are and the file still opens in a spreadsheet. A redacted lead export keeps its shape for testing an import or showing a colleague the layout, without real contact details.
Is redacted text anonymous under UK GDPR?
Not necessarily. Removing direct identifiers reduces risk, but someone may still be identifiable from what remains, such as a job title at a small firm in a named town. The ICO's guidance on anonymisation explains the test. Treat redacted text as personal data unless you are sure nobody could be identified.
Where your data goes
Redaction runs in this tab, with large texts handed to a background worker in the same browser. The original text never leaves your device, and nothing is stored once you close the page.
Checked against the tool on