Security, Abuse & Privacy

Personal Data

Definition

Under the GDPR, personal data means any information relating to an identified or identifiable natural person. The word identifiable is doing the heavy lifting: it covers data that names someone outright and data that singles them out when set beside information someone else holds. So a name, an email address, a photograph and a phone number are obvious members, and so are cookie identifiers, device fingerprints, order numbers and login handles. A special category sits on top — health, ethnicity, religion, political opinions, trade union membership, sexual orientation, biometrics — with stricter conditions attached. Pseudonymised data, where names are replaced by reference codes but a key still exists somewhere, remains personal data; only genuinely anonymous data, which nobody can reverse, falls outside the rules entirely, and that bar is higher than it sounds.

Why It Matters

The IP address case is the one that catches website owners out. In Breyer, decided by the Court of Justice in 2016, a dynamic address held by a website operator was found to be personal data, because the operator could plausibly identify the subscriber with the help of the internet provider. That single conclusion is why cookie-free analytics still has thinking to do: the banner may be unnecessary, but the addresses arriving in every request are regulated the moment you write them down. The usual answer is not to write them down — derive a country and a repeat-visit signal, then discard, or hash the address with a salt that rotates each day so yesterday's visitors cannot be matched to today's. Nothing here is legal advice, but the practical rule holds: data you never stored cannot leak, cannot be subpoenaed, and cannot appear in a subject access request.

How It Works

Working out whether something is personal data is a test of what could be linked, not of what the field is called. Start with the records you actually hold — form submissions, access logs, support emails, billing rows, the analytics store — and ask, for each, whether one person could be picked out from it alone or with reasonable help. Where the answer is yes, record the purpose, the lawful basis and how long you keep it. Reduction is the main technique available to a small operation: truncating an address to its network part, hashing with a rotating salt, keeping daily totals rather than the rows they came from, and deleting the raw material once the aggregate exists. Be honest about one thing, though: hashing alone is weak on a small set of possible values, because anyone with the salt can hash all four billion IPv4 addresses and match them back, which is exactly why the salt has to rotate and never be stored beside the hashes.

Real-World Example

A recruiter publishes a hiring pack at roles-2026.99helpers.site with a form for candidates to register interest, collecting name, email and a short note. That form store is plainly personal data and gets a purpose, a notice and a ninety-day life. The visit statistics beside it are not the same thing: 99helpers counts views, referrers and countries at the edge without storing raw addresses, so the traffic report is a set of numbers rather than a record of who looked at the role. Two stores, two very different obligations, and telling them apart took one conversation.

Common Mistakes

  • Believing that removing names makes data anonymous — if a key or a linkable identifier still exists, it is pseudonymous and still regulated
  • Filing raw addresses in an analytics table for future analysis — the future analysis never happens and the liability accumulates anyway
  • Copying special category data without noticing, such as a dietary requirement field on an event form that reveals religion or health

Related Terms

Put a file online in seconds

Drop in a document, an image, a page or a whole static website and share the link — free, with no build step and no server to set up.

Host a file free →