Fix a document that was read wrong
Sometimes Merchaint reads a document as the wrong type, or misses a value. Reprocessing gives it another go — and lets you tell it which schema to use, so it doesn't make the same mistake twice.
- A walkthrough video, then why reprocessing is the fix
- Why there are two separate switches, and which one costs you
- Use cases for each kind of problem, and which settings each one needs
- The steps to reprocess a document, then troubleshooting notes and FAQ
01 · Video
Feature walkthrough
02 · Why
Why do we have this?
A document read as the wrong type produces the wrong fields, and those fields flow into matching, approvals and your ledger. Fixing it at the document is the cheapest point in the whole chain.
Why are there two separate switches?
Because there are two different problems. If the values were read correctly but filed in the wrong place, re-filing is enough and costs nothing. If the values themselves are wrong, the document has to be read again, which does cost.
Keeping them separate means you never pay for AI you didn't need.
03 · Use cases
A credit note read as an invoice. Reprocess and pick the credit note schema.
One supplier's invoices keep landing as the wrong type. Reprocess each one with the right schema, and tell your admin — the schema may need changing.
The values are wrong, not misfiled. Reprocess with Force re-extract on. If it's still wrong, the image is the problem — scan it again.
The document page shows an amber banner: "Not mapped to a schema — this document's data isn't filed into a table yet." It was read, but no records were made. Choose schema on that banner opens the same box as Reprocess.
Which settings do I want?
| Your problem | Schema | Force re-extract |
|---|---|---|
| Right values, wrong type | Pick the right schema | Off |
| Wrong values, right type | Leave on Auto-detect Schema | On |
| Wrong values and wrong type | Pick the right schema | On |
| Read, but never filed | Pick the right schema | Off |
| No schema fits at all | Create a new schema | On |
04 · How
- Open the document from
Inboxand work out what's actually wrong — the type, the values, both, or that it was never filed. - Click
Reprocess— orChoose schemaif the not-mapped banner is showing. They open the same box. Leave it onAuto-detect Schemato let Merchaint try again, or pick the schema it should have used. - Set
Force re-extractusing the table above. - Confirm, then check the result once it's back to
Ready.
05 · Expected result
The document goes back to processing and returns to Ready. The type now matches what you chose, and the fields are the ones that schema looks for. If you re-read it, the values should now match the page.
06 · Troubleshooting / notes
| What you see | Why | What to do |
|---|---|---|
| Reprocessing changed nothing | Force re-extract was off, but the values were the problem | Do it again with it on |
| Still wrong after re-reading | The document isn't clear enough to read | Scan or photograph it again |
| No schema in the list fits | This document type isn't set up yet | Create a new schema, or ask your admin |
| The same supplier keeps going wrong | A setup problem, not a one-off | Tell your admin — see Schema Studio |
07 · FAQ
Will this cost me anything?
Only if you turn on Force re-extract. Left off, reprocessing reuses what was already read and costs nothing.
Can I fix one field instead of the whole document?
This article covers reprocessing. If one value is wrong and everything else is right, ask your admin whether it can be corrected on the record.
Common questions
That's a schema problem, not a document problem. Your admin can change the schema in Schema Studio.