PII masking & privacy | ATX AnalyticsTable Handbook

PII masking & privacyFour methods, a detector that shows its evidence, and honest limits. Mask PII replaces personal data in chosen columns or a chosen range with something safe to keep looking at, so an extract needed for analysis, a demo or a screenshot can leave the machine without the people in it leaving with it. It works on grid data, on static pinned data, and as a pipeline step that re-masks on every run. Masking is one-way by design: there is deliberately no unmask button and no stored map back to the originals, because a project file holding that map would contain precisely the data it appears to have removed.

What it does

Four methods — Redact (fixed-width, so length leaks nothing), Partial (keep a recognisable fragment), Hash (a stable token that still joins), Synthetic (a plausible fake of the same shape, with valid checksums)
Scope by whole columns or by a range, with header-row detection so several tables in the same columns stay independent
A detector that scans the rows in scope and labels each suggestion 'by value' (format and checksum verified) or 'by name' (column heading only)
Recognises e-mail, card (Luhn), IBAN, PESEL, postcode, phone, names, addresses, birth dates and customer identifiers — with Polish and English column headings
A per-project 128-bit key makes hash and synthetic repeatable across sheets and scheduled runs, and different between projects
Available as a pipeline step that masks in place, arrives un-scanned, and fails loudly if a named column is missing

How it works

Open the dialog on a selection; it scans the rows in scope and pre-ticks what it recognises, always proposing Redact first because the safest method should need no decision.
Values outrank names: a column called 'notes' full of e-mail addresses is treated as e-mail, and a column called 'email' full of order references is not.
Hash keeps equality, so distinct-customer counts, groupings and joins still work with no real identifier in the file.
Synthetic produces checksum-valid IBANs, cards and PESELs and RFC-2606 example.com addresses, which is what makes a masked file usable as a UAT extract.
The pipeline step refuses to run without a project key rather than falling back to a constant — masking under a known constant looks exactly like masking and protects nothing.

Practices that keep it reliable

Keep an unmasked copy of anything still needed before you mask; once the project is saved the originals are gone
Send the masked extract, not the project — the project file carries the key
Match the method to the purpose: redact for a regional report, hash for a UAT extract, synthetic for a demo, partial for a call-centre view

In a regulated environment

Masking is not anonymisation. Re-identification works on combinations — a postcode, a birth date and a branch can still name a person — and this is a property of the whole table, not of one column.
Masking is not encryption and not deletion: under GDPR, masked data derived from personal data may still be personal data if re-identification remains possible.
The pipeline file never carries the key, so a masking recipe can be committed, shared and reviewed without exposure.

Screens

Mask personal data — scope, header handling, a per-column method and a preview of the result.

Related guides

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.