Home · Free tools · Invoice data extraction
Invoice data extraction, without buying an AP platform
Two honest notes first. Most of what ranks for this phrase is enterprise accounts-payable software — a platform your finance team buys, connects to an ERP, and runs thousands of invoices through. TableDI 2 is not that, and this page will not pretend otherwise. Second: half of what people call extraction is not extraction at all. If your invoice PDF has real text in it, the table can be pulled out right here in this tab, with nothing uploaded and nothing guessed — try it below. A scan or a photo genuinely needs OCR, which neither this page nor TableDI 2 does; the rest of the page is the map for that case.
Open a PDF invoice
Drop a .pdf here, or click to choose one
Doing this every month? Do it once in TableDI 2, save it as a job, and run it again next month. Download TableDI 2 free
Nothing is uploaded. The file is read by JavaScript in this tab and never leaves your computer — no server sees it, and closing the tab throws it away.
Turning a PDF invoice into a spreadsheet
- Check whether it is text or a scan. Open the PDF and try to select the text with your cursor. If it highlights, there is a text layer and no model is needed.
- Text PDF — pull the table out above. Drop it into the box on this page. The rows are read by JavaScript in this tab, pick the page, and download CSV or .xlsx. Nothing is uploaded, and no value is guessed.
- Scan or photo — it needs OCR. A scanned page has no text to read. Neither this page nor TableDI 2 reads scans: ask the supplier for the original PDF or a structured file, or use an OCR service you are willing to send the document to.
- Reconcile before it reaches your books. Check the extracted total against the document. Line-item tax is the most common silent failure, whatever tool produced the numbers.
Three routes, and who each one is for
| Route | Fits when | Where the invoice goes |
|---|---|---|
| Enterprise AP automation Nanonets, Affinda, Stampli, DocuClipper, Klearstack |
Hundreds or thousands of invoices a month, approval workflows, an ERP to post into | Their cloud, under a contract your company signed |
| A free online converter | One invoice, once, and you do not mind who sees it | Uploaded to their server; retention policy varies and is often unstated |
| This page text PDFs only |
Invoices that arrive as real-text PDFs, a person rather than a department, and data you would rather keep to yourself | Nowhere — the PDF is read by JavaScript in this tab |
The middle row is the one worth thinking about twice. An invoice carries a supplier name, bank details, line-item pricing and often a contract reference. Posting that to a free service is a disclosure of commercial terms, and free converters differ enormously in how long they keep uploads.
What "extraction" actually has to get right
Invoice layouts are the hard case in document extraction, and it is worth knowing where any tool — ours included — goes wrong:
- Line items versus totals. Pulling the invoice total is easy. Pulling every line with its quantity, unit price and tax code is the part that separates tools.
- Tax. Several rates on one invoice, tax shown per line or only in the summary, reverse charge. This is where numbers silently stop adding up.
- Dates. Invoice date, due date and delivery date all appear; 03/04/2026 means two different things depending on the country the supplier is in.
- Multi-page invoices where the table continues across a page break with a repeated header.
Whatever you use, reconcile the extracted total against the document before anything reaches your books. That check is not optional with any tool at any price.
When the invoices come back every month
If the invoices already live in your accounting system, its export — a CSV or Excel list of invoices — is a better source than any PDF. TableDI 2 is a desktop app for file work you redo every period: it keeps your sources, rules and delivery as a job, so next month you drop in the new files and run it again. Everything runs on your own machine. Matching that invoice list against the payments on your bank statement is one of the jobs it is built for: the rows that do not match come back as exceptions, each showing the file and row it came from.
For thousands of invoices a month, with approvals and posting into an ERP, an AP platform is the right purchase and we would tell you so.
Getting the extracted data into Excel
Most people who search for this want one thing at the end: a spreadsheet. Which route gets you there depends on what the invoice is made of, not on which tool is best:
- A text PDF (the text highlights when you select it) — the box at the top of this page reads the line-item table out of it and hands back CSV or .xlsx, in this tab, with nothing uploaded and no value guessed. For bank and card statements in the same shape, the bank statement converter knows the debit/credit columns and checks the running balance.
- A scan or a photo — this is the case that needs OCR, which neither this page nor TableDI 2 does. Ask for the original first; failing that, use an OCR service you are comfortable sending the document to, and check the totals afterward.
- A structured file from the supplier (CSV, UBL, Factur-X) — open it directly. It is more accurate than any extraction, because nothing was read off a page.
Whichever route, the Excel file is the ordinary kind: no add-in, no link back to a service, and it opens the same way in Excel, Numbers and Google Sheets.
Before you extract anything, check whether you have to
Two shortcuts that beat every extraction tool when they are available: many suppliers will send invoices as a structured file (UBL, Factur-X, ZUGFeRD or plain CSV) if you ask, and a PDF whose text is real text needs no model at all — select the text in a viewer, and if it highlights, it can be copied. Only a scanned page genuinely needs extraction.
Questions people ask
What is the cheapest way to extract data from invoices?
For a small number: ask the supplier for a structured file, or copy the text out of a text-based PDF — both are free and more accurate than any model, because nothing is being guessed. For a steady trickle of scans, an OCR service priced per page is the usual route. For thousands a month, an AP platform is cheaper per document than anything else.
Does TableDI upload my documents?
No. The tools on this site read the file with JavaScript in your tab, and TableDI 2, the desktop app, reads it on your own machine. The one exception is optional: if you add an AI key in TableDI 2, drafting a job sends the column names and the first 50 rows straight from your machine to the AI provider you chose — never the whole file, and never through anything of ours.
How accurate is invoice data extraction?
Good on clean, consistent layouts; unreliable on faint scans, phone photos at an angle, and multi-page tables with repeated headers. Line-item tax is the most common silent failure. Reconcile the extracted total against the document every time, whatever tool you use.
Can it read PDF invoices, or only images?
PDFs with a text layer — that is what the box at the top of this page reads, in your browser. Images and scanned PDFs need OCR, which this page does not do. Check which kind you have by trying to select the text in a viewer first.
How do I get invoice data extraction to Excel without an AP platform?
For a text PDF, drop it into the box at the top of this page and download .xlsx — nothing is uploaded and no value is guessed. A scanned invoice needs OCR first, which this page does not do. If what you have is a bank or card statement rather than an invoice, the bank statement converter does the whole job in the browser.
The converter on this page says my PDF has no text layer. Now what?
That means the file is a scan or a photo — a picture of an invoice, with no characters in it to read. Nothing on this page can help, and any tool that claims to needs OCR or a model. Two routes: ask the supplier for the original PDF or a structured file, which is both free and exact; or run it through an OCR service you trust with the document, and check the totals against the invoice.
Is this an accounts payable system?
No. There is no approval workflow, no ERP posting, no supplier portal and no audit trail. This page turns a text PDF into a table, and TableDI 2 reconciles files like that against each other, period after period. If you need AP automation, the platforms named above are built for it.
Related
Doing this every month?
In TableDI 2 you do it once, then save it as a job. Next month you drop in the new files and run it again.
macOS, Apple silicon and Intel; Windows is in progress (what to do meanwhile). Free is not a trial — no account, no card.
Last reviewed 2026-09-11 by the TableDI team. Something wrong on this page? Tell us — it is one inbox, read by the people who build TableDI.