Provenance is who, when, from where.

Open any agency's shared drive and you will find the same thing: folders named after students, full of files named after nothing. Scan_0043.pdf. passport_final_v2.jpg. transcript.pdf. Each of these was, at some point, sent by somebody, for some purpose, and checked by somebody or not. None of that survives in the file.

Provenance is the missing half. It is the answer to four questions about every document you hold: who supplied it, when, through what channel, and what has happened to it since.

The four facts.

Who supplied it.

Not "the student" as a category — the actual party. The student, a parent, a sub-agent, the institution, an examination body, a translator. The difference matters because the weight you can put on a document depends on how far it is from its issuer. A transcript emailed by the issuing university is a different object from the same transcript forwarded by a partner who received it from a family friend.

When.

The moment it entered your custody, recorded automatically rather than typed. Many documents have a shelf life — bank statements, police certificates, medical results, test scores. An upload timestamp is what lets a system tell you a document has gone stale before an institution does.

Through what channel.

Uploaded by the student to their own portal, emailed to a counsellor, handed over on paper and scanned in the office, pulled from an institution's system. The channel is most of what a reviewer needs to judge how a document was handled, and it is the fact most often lost, because a file dragged into a folder from an inbox looks identical to one uploaded by the student.

What has happened since.

Replaced by a newer version, superseded by an original, checked against the issuer, marked verified by a named person, or attached to a submitted application. A document's history is often more informative than the document.

Provenance is not the same as verification.

These get conflated constantly and they answer different questions. Provenance says where a document came from. Verification says somebody checked it against its issuer and how. A document can have excellent provenance and still be a forgery the student was given in good faith. It can be genuine and have no provenance at all, which means you cannot demonstrate anything about how you handled it.

Keep them as separate facts. Marking a document verified should require a named person and a stated method, and it should be possible for a document to sit unverified indefinitely without that looking like an omission — because for most documents, it is the correct state.

What it buys you.

• An institution asks how a qualification was assessed. You answer in one screen instead of searching three inboxes.
• A family disputes what was submitted on their behalf. You can show exactly which version of which document went with which application, and when.
• A staff member leaves. Their students' files are still legible, because the context was in the record rather than in their memory.
• A document expires. The system knows, because it knows when the document arrived and what kind of document it is.
• A sub-agent's files turn out to need rework more often than anyone else's. You can see that as a number, because every document carries its source.

Making it survive contact with a busy office.

Provenance fails when it is extra work. Three design decisions keep it honest.

1. 01Capture the four facts automatically at upload. Nobody types them, so nobody skips them.
2. 02Make replacing a document an explicit action that keeps the old version, rather than an overwrite. The history of a document is evidence; a silent overwrite destroys it.
3. 03Give students and sub-agents their own upload route. A file that arrives through a channel the system understands carries its provenance for free; one that arrives as an email attachment does not.

The prize is not tidiness. It is being able to answer a hard question about something that happened eighteen months ago, quickly, without asking anyone to remember.