Your bank statement is a scan, and that is why nothing happened

Two PDFs can look identical on screen and be completely different underneath. One carries a text layer: the characters are really there, and any program can read them exactly as the bank wrote them. The other is a picture of a page — a scan, a photo, or a printout that was scanned back in — and contains no characters at all, only pixels arranged to look like them.

Drop the second kind in here and you are told plainly: this is a scan or a photo, with no text to read. It cannot be converted, and nothing has been sent anywhere.

How to tell before you upload

• Open the PDF and try to select a line of text with your mouse. If the text highlights word by word, there is a text layer. If you get a blue box over the whole page, or nothing at all, it is an image.
• Try your reader’s find function and search for a payee you can see on the page. No match means no text.
• Statements downloaded from online banking almost always have a text layer. Anything that has been through a scanner or a phone camera almost never does.

Why refusing beats guessing

Reading a scan means text recognition, which turns pixels into characters by estimating. That is a reasonable thing to do, but it produces a different kind of output: figures that were inferred rather than read. The whole point of the check performed here is that every row is re-derived from the balance the statement prints — and a recognised figure can be wrong in exactly the way that check exists to catch, while looking perfectly ordinary.

Text recognition is not built here, and it would have to run on a server rather than in your browser, which changes the privacy position entirely. Rather than do that quietly, the file is refused and nothing leaves the page.

Where to get a readable copy

1. Log in to online banking and download the statement as a PDF rather than scanning the paper copy. Most banks keep at least a year, often several, and the downloaded file will have a text layer.
2. Look for a CSV, Excel or OFX export while you are there. If your bank offers one, it skips conversion altogether.
3. Ask the bank for a duplicate e-statement if the period you need has dropped off the site. They can usually reissue it.
4. If the only copy that exists is paper — an old account, a closed one, a probate file — recognition somewhere else is the only route, and the figures it produces should be checked by hand against the paper before you rely on them.

When only some pages are scans

Mixed files turn up more than you would expect: a statement downloaded properly, with one page re-scanned after being signed or stamped. Those pages are counted and named, the readable pages are converted, and the scanned pages are left out rather than silently skipped. Treat the result as incomplete: the transactions on the missing pages are absent, so the totals will not match the statement.

If a file looks like an ordinary downloaded statement and is still refused, that is worth telling us about. The results screen offers a diagnostic report describing the layout — page count, where the columns sit, what was recognised — with no amounts, dates or payee names in it. Open it, read it, and send that rather than the statement.