Define one row before defining the columns
Write a sentence explaining what a row represents. If the sentence includes both an invoice and a product, you probably need two sheets. Keep the invoice total on the document sheet and connect each item to that document. Otherwise, a three-line invoice can accidentally contribute its full total three times when someone sums the item sheet.
Choose a small set of document fields
Capture only fields needed for the receiving workflow. Preserve the source's invoice number as text, including letters, punctuation and zeros. A supplier's number may repeat elsewhere, so use your own source reference as the connection between sheets. Record the printed date separately from the date your team received the file.
- Source reference and supplier reference
- Document type: invoice, credit note or unresolved
- Printed invoice number and printed document date
- Currency as stated, printed subtotal, tax and total where required
- Exception status and the source location of a question
Keep line-item meaning intact
A useful item sheet can include source reference, page or line reference, product code, description, quantity, unit and printed amount. Include pack size when it changes the meaning of quantity. A quantity of two cases is not automatically two individual units. If the source has discounts or multi-line descriptions, decide their representation before the full batch is entered.
Separate captured values from calculated values
A printed total is evidence of what the document says. A calculated total is a separate check with its own assumptions. Label them accordingly. Do not invent a tax rate, convert currency or infer payment status to fill a convenient column. Credit notes should retain their document type and stated sign convention; the authorized bookkeeper decides how to use them.
Agree the exception vocabulary
Use distinct states for not shown, unreadable and awaiting clarification. A blank numeric field must not silently become zero. Test the map on a small authorized sample covering the main layouts, including one awkward document. Check that another person can trace an uncertain cell to its source without searching the whole collection.
Source reference: INV-SAMPLE-014 Row unit: one printed line item Quantity source: page 1, item 2 Unit: case Pack size: not shown Question: confirm pack size before unit comparison
Use the map to scope the work
Your field map, approximate volume and supplier-layout count are enough to start a scope conversation. It is not necessary to paste invoice contents into a public form. Detailed source review belongs in an agreed private channel. Define whether delivery includes summaries, item rows, comparison checks or all three, and name who will resolve exceptions.
A few useful answers
Should every invoice have the same fields?
Use one agreed output schema, but keep unavailable values explicit. A consistent schema does not mean every supplier provides every field.
Can invoice numbers be the unique identifier?
They may not be unique across suppliers or document types. A stable internal source reference avoids relying on that assumption.
Does an extracted invoice count as reconciled?
No. Extraction prepares source data. Reconciliation, accounting treatment, approval and payment remain separate professional or business decisions.