PDF invoice to CSV: columns first, layout never
PDF to CSV works when you map fields, not positions: vendor, number, date, net, tax, total. Position-based conversion breaks as soon as the layout changes.
5 documents free. PDF, photo or XML — data out as CSV, Excel or JSON.
Why layout-based extraction breaks
A PDF has no columns — only text placed at coordinates. Change the template and the coordinates move.
The CSV then silently shifts values between columns, and the totals look plausible.
The mapping that holds
Define the fields you need once: vendor, invoice number, dates, net, tax, gross, currency.
Extraction fills those fields; the CSV is written from them.
Batch conversion
Many PDFs at once produce one CSV with a row per document plus a failure list.
Duplicate key: vendor plus invoice number.
Checks before import
Sum of tax equals invoice tax.
Row count equals file count minus failures.
Frequently asked
Does it work with scanned PDFs?
Those go through recognition first; the CSV comes out the same shape.
Can I choose the delimiter?
Yes — CSV is written for the tool you import into.