Batch invoice processing: volume is not the hard part
Batch processing means many files at once, one row each. The hard part is not speed — it is keeping totals right and catching duplicates when the same invoice arrives twice.
5 documents free. PDF, photo or XML — data out as CSV, Excel or JSON.
Why volume changes the problem
One invoice: you check it. A hundred: you check the pattern, not the document.
So the checks must be automatic: totals, tax per rate, duplicates.
What a batch run returns
One sheet, one row per document, plus a list of files that failed and why.
Failures are normal: a blurry photo, a password-protected PDF, a file that is not an invoice at all.
Duplicate detection
Vendor plus invoice number is the reliable key. File names are not — the same invoice arrives as scan1.pdf and INV-2291.pdf.
Flag duplicates for review rather than dropping them silently.
Before you import
Check the count: rows should equal files minus failures.
Check the totals: the sum of a sample against the source documents.
Frequently asked
How many files can I process at once?
Batches are handled sequentially with a limit on parallelism; large runs take longer but do not fail.
What happens to files that fail?
They are listed with a reason, so you can fix the input instead of guessing.