Skip to content
Account

How to extract a table from a scanned PDF to Excel

A scanned table has to be read with OCR and then rebuilt as rows and columns from where the text sits on the page. Use a converter made for scans, download the spreadsheet, and check the cells it marks as uncertain. Where a column should add up — an invoice total, a running balance — check the sum: a number read from a picture can be wrong.

Updated 1 October 2026

Extract a scanned table to Excel

  1. Pick the converter for your documentA scanned invoice: Scanned invoice to Excel. A scanned bank statement or passbook page: Scanned bank statement to Excel. Any other table in a picture: Screenshot to table.
  2. If your table is in a scanned PDF and is not an invoice or statementTurn the page into a picture first with the free PDF to JPG (choose 300 dpi for small print), then give the picture to Screenshot to table.
  3. Run it and check the marked cellsEach costs 1 credit a page or picture, shown before it runs. In the result, the cells the reader was least sure about are highlighted; check those against the scan.
  4. Download and check the totalsDownload the Excel file (CSV and JSON are offered too). Add up any column that should match a printed total.

Why scanned tables are harder than digital ones

In a PDF made by software, each number is stored as text at a position, so a converter can read it exactly and only has to work out the grid. In a scan, every character has to be recognised from pixels first. A blurred digit, a faint ruled line or a slight skew can move a value to the wrong cell or change a 5 into a 6.

So a good scanned-table converter does three things a plain one does not: it straightens the page, it marks the cells it is unsure of, and — where a column is meant to add up — it checks the sum and reports a mismatch instead of quietly correcting it.

Is your PDF scanned?

Try to select a number in the table. If you can, the PDF has a text layer and the cheaper, exact route works: PDF to Excel, Invoice PDF to Excel or Bank statement to Excel. If you cannot, use a scanned converter.

Honest limits

What happens to the file

The scan is uploaded for the conversion and deleted from our servers after 2 hours, whether or not you download the result. It is sent to the OCR provider and the AI provider that read it, which do not train on it.

Questions

Why can't I just copy the table out of the scan?

Because there is nothing to copy: a scanned page is a picture, with no text in it. The page has to be read with OCR first, and the grid rebuilt from where the text sits, which is what a scanned converter does.

Is there a free way?

For a few rows, the free Image to text reads the text out of a picture in your browser, and you can paste it into Excel and split it into columns yourself. The paid converters rebuild the grid for you and mark uncertain cells.

How accurate is it?

We do not publish an accuracy figure yet. It depends on the scan: a sharp, straight 300 dpi print reads far better than a blurred phone photo taken at an angle. That is why uncertain cells are marked and totals are checked, so you know which cells to look at.

Do you have a scanned table to test with?

Yes. The sample scanned invoice has the same numbers as the digital sample invoice, so you can compare what a converter gets with the right answer. The three-page sample scan has two tables whose columns add up.

Tools for this

Free sample files to try it on

More guides