Verifying an export is actually complete

The export arrived and the file is on your disk. That is not yet evidence of anything: an archive can download cleanly, open cleanly, and still be missing half of what you assumed was in it.

The check takes about fifteen minutes and has to happen before you cancel anything, because the only tool that can produce a replacement is an account that still exists.

What can go wrong without looking wrong

The failure mode is silent partiality, not corruption. A truncated archive is not empty. It opens, it contains conversations, the conversations look right — and the oldest six months are absent, or attachments resolve to nothing, or one export scope was selected and another assumed.

Nothing warns you. There is no completeness indicator on a file, and the export itself has no idea what you expected. The verification exists because the artifact cannot self-report.

Check one — does it open at all

Extract the archive fully rather than previewing inside the compressed file, then open whatever the top-level entry point is with something other than the tool you happened to have handy.

If the archive contains a browsable file, open it in a browser. If it contains structured data files, open one in a plain text editor and confirm it is readable text rather than a placeholder. A partial extraction that stops halfway is common with large archives and looks like success until you notice the tail is missing.

Check two — find the oldest thing you remember

This is the check that catches range problems, and it is the one people skip.

Think of a specific conversation from as early in the account’s life as you can remember — a recognisable one, with a phrase you could search for. Find it in the archive.

If the earliest content in the file is from a few months ago and the account is two years old, the export is scoped, partial, or still processing, and none of those is visible from the file size. Do this before assuming the whole history is there, because “the whole history” is exactly the assumption an export invites.

Check three — find the newest thing

Same procedure at the other end. The last substantive conversation before you requested the export should be present.

If it is not, the snapshot was taken earlier than you think, or the request was queued for longer than you think. Either way the snapshot boundary is not where you assumed, and every plan that depends on the boundary needs adjusting.

Check four — attachments and uploads

Pick three conversations where you uploaded a file and follow the reference.

Attachments may be included as actual files, referenced by name only, or absent entirely, and which of those is true varies by product and changes over time. It is not something to assume in either direction — check, then act on the answer.

If files are absent, the source documents matter more than the archive does, and collecting them separately becomes a step of its own rather than something the export handles.

Check five — everything that is not conversations

An account is more than its chat log. Look through the archive for the standing instructions, any saved personalisation or memory, saved prompts, and the organisational structure.

Some of these appear in some exports. Assume none of them are present until you have found them, because the whole category of things that do not transfer is easiest to plan for while the source account is still open. If memory is not in the file, copy it out by hand before anything gets cleared.

What moves

WHAT MOVES — verification

  · An archive that opens fully
                    → confirmed readable. This is the
                      only claim a successful extract
                      supports.

  · Conversation range, oldest to newest
                    → confirmed only by finding both
                      ends yourself.

  · Attachments
                    → varies. Present, name-only, or
                      STAYS BEHIND entirely. Check three
                      before deciding.

  · Instructions, memory, saved prompts
                    → often STAYS BEHIND. Treat absence
                      as the default.

  · A failed export found after cancelling
                    → IRREVERSIBLE. The account that
                      could reissue it no longer exists.

Write down what you found

Two or three lines, kept with the archive: the date of the snapshot, what is in it, and what is not.

This is not bookkeeping for its own sake. In two years you will open this folder looking for something, and the note is what tells you whether the thing was never exported or whether you are looking in the wrong place. Without it, an incomplete archive is indistinguishable from a bad search, and you will spend an hour on the wrong one.

Keep the note in plain text alongside the files, not in a separate notes app, so it survives with the archive rather than separately from it.

If a check fails

Request a new export while the account is active, and change one thing at a time: check the export scope options, wait longer before downloading, extract with a different tool.

If it fails repeatedly and the missing material matters, the fallback is manual. Open the conversations you cannot afford to lose and copy the text out yourself. Tedious, reliable, and available only while the account is open — which is the reason verification comes before cancellation rather than after.

What verification cannot tell you

Whether the archive is useful. A complete export of two years of conversations is a large pile of text that nothing indexes, and completeness is not the same as being able to find anything in it. That is a separate job with a separate afternoon attached.

And it cannot recover the gap between the snapshot and the day you stopped. Verification confirms what is in the file. It has nothing to say about what was never put there.