Customer Onboarding Dataset Anonymisation with anonym.plus

Turn an onboarding dataset into truly anonymous records on your own machine.

In simple terms, PII redaction is the on-device process of finding and masking personally identifiable information in a document before it is shared.

UK GDPR Recital 26 draws the line that matters here: data protection law does not apply to anonymous information, and identifiability is judged by the means reasonably likely to be used by anyone. DPA 2018 s.3(2) states the same test in statute. Art. 4(5) puts pseudonymised records firmly on the other side of that line — if a key exists, they are still personal data. The ICO tests the result with a motivated intruder: could a reasonably competent person with no special skills re-identify someone? DPA 2018 s.171 then makes it an offence to re-identify de-identified personal data without consent. anonym.plus strips each identifier on your machine, and turning the name map off is what makes the result one-way.

When this applies

An onboarding export is going to train a model. Names, National Insurance numbers, IBANs and contact rows would be learned along with everything else. You de-identify the set first, with the name map off, so no row links back to a person.

How anonym.plus handles it

  1. Point anonym.plus at the export on your machine.
  2. It scans every column for the identifier set.
  3. Local OCR reads any scanned rows in the mix.
  4. Turn OFF the name map so the result is one-way.
  5. Apply Replace or Redact across the dataset.
  6. Save the anonymous set locally.

What you need to provide

PII & financial identifiers detected

Categoryanonym.plus entity typeExample
NamesPERSONCarla Davies → [PERSON]
IdentifiersUK_NINOQQ 41 29 08 C → [NINO]
FinancialIBAN_CODEGB29 NWBK 6016 1331 9268 19 → [IBAN]
ContactEMAIL_ADDRESSdavies@example.co.uk → [EMAIL]
FinancialUK_SORT_CODE60-16-13 → [SORT_CODE]
DatesDATE_TIMEDOB 1982 → [DOB]

Compliance achieved

Anonymise customer-onboarding datasets offline — see plans & start free →

Limitations & cautions

True anonymity is hard. A rare combination of age, branch and amount can still single out a person, which is exactly what the ICO's motivated-intruder test is designed to expose. Removing direct identifiers is a start, not a guarantee — assess re-identification risk before release.

Frequently asked questions

What makes a dataset anonymous under UK GDPR?

Recital 26 treats information as anonymous only where nobody can re-identify a person by any means reasonably likely to be used. The ICO checks that with a motivated-intruder test. Turn the name map off so no key links back, then assess what the remaining columns could still reveal in combination.

Is pseudonymisation the same as anonymisation?

No. UK GDPR Art. 4(5) describes pseudonymisation as processing where the data can still be attributed using separately held information. A reversible name map is exactly that, and the output stays personal data. For Recital 26 anonymity, keep the map off.

What happens if someone re-identifies the set later?

DPA 2018 s.171 makes it an offence to re-identify de-identified personal data without the consent of the controller who de-identified it. That is a real constraint on a recipient, but it is not a substitute for doing the work properly — the offence does not make a weak dataset safe.