How to Convert a PDF to Excel Without Uploading It

A bank statement, an invoice, a supplier price list — the numbers you need are right there, and none of them are selectable. Here's why Excel's built-in option leaves most people out, why the free online converters are the wrong answer for financial documents, and how to get clean rows without either.

Last updated: September 2026

The two usual answers, and what's wrong with each

Excel does ship a PDF importer: Data → Get Data → From File → From PDF. It works, and if it's available to you it's a reasonable first try. But it carries two limits that catch people out. It is Windows-only — it runs through Power Query, so it simply isn't there on Mac or Excel for the web. And it can only find tables that already exist as text; hand it a scanned document and it returns nothing at all, because there is no text for it to find.

So most people reach for a free online converter instead. Those work well enough — but every one of them requires you to upload the document to a stranger's server. For a marketing PDF, fine. For a bank statement, a payroll summary, a signed contract or a client invoice, you have just handed a third party a confidential financial record, and in many workplaces that is a policy breach regardless of how the file is handled afterwards.

That leaves the third answer most people settle on, which is retyping — slow, and the one approach guaranteed to introduce errors into figures that need to be right.

Read, not recognised — why that distinction matters

It's worth being precise about what a "PDF converter" is actually doing, because the two approaches have very different failure modes.

A digital PDF — one produced by a bank, an accounting system or a Save-as-PDF — already stores every character along with its exact position on the page. Reading a table out of it is a matter of recovering the layout, not identifying the characters. Nothing is guessed, so a 3 can never come back as an 8.

OCR is a genuinely different operation: it looks at an image of a page and infers which characters the shapes represent. It is remarkable technology and it is also, unavoidably, probabilistic — which is exactly why it should be the fallback rather than the default. Many converters run OCR on everything, discarding perfect data to re-recognise it. Tellsheet uses it only where there is no alternative, and tells you which pages it was used on so you know which rows to spot-check.

Things worth knowing

Up to 50 pages are read in one go; for a longer document, give a page range like 2-7 in the Pages box. If a page's text layer is present but scrambled — which a few PDF generators do produce — switch on Force OCR to read it as an image instead. And if a page yields nothing at all, Tellsheet says so by page number rather than silently returning a short table.

One honest limitation: a table with no visible column gap — where one cell's text runs straight into the next — can lose that column boundary, because the gap is what marks the split. Tables with normal spacing, which is nearly all of them, come through cleanly.

Frequently asked questions

Related Excel guides