The task
A supplier statement rarely agrees with your ledger on the first pass, and most of the gap is timing rather than error. Configure an agent to line the two records up, group the differences by the reason they exist and hand the remainder to a person. What you get back is a working list, not a verdict.
What you provide
Supplier statement
Statement ST-0907 from a supplier, listing document numbers, dates and amounts, often as a PDF or a scan.
Your open items
An extract of open items for that supplier from your ledger, with document numbers, posting dates and amounts.
Your reconciliation rules
How you treat goods in transit, credit notes, part payments and documents posted after the statement date.
Give the agent a clear brief
Use these instructions as a starting point, alongside your approved sources and permissions.
Align the two records on document number, date and amount. Group every difference by reason: present on the statement only, present in the ledger only, a different amount, or a timing difference under the supplied cut-off rules. Cite both source lines for every difference. Do not adjust the ledger, post a correction or contact the supplier.
How the work comes together
Read both records
Extract the statement lines and normalize document numbers, dates and amounts against your ledger extract. Keep the page reference on the statement so a reviewer can find the original line.
Group the differences
Separate timing differences from genuine gaps. A document posted after the statement date is an explanation, not an exception, and it should never sit in the same list as an amount that disagrees.
Hand over the remainder
Return the unexplained differences with both source lines and a running total. That total is the amount your team actually has to resolve with the supplier.
Illustrative example · fictional data
Most of the gap explains itself
An illustrative reconciliation of a fictional supplier statement against a ledger extract, using a statement cut-off of the last day of the month.
| Difference | Both sources | Reason |
|---|---|---|
| Statement only, $4,200 | ST-0907 line 6 · no ledger entry | Goods in transit at cut-off |
| Ledger only, $1,150 | Your entry 44821 · not on ST-0907 | Credit note raised after the statement date |
| Amount differs, $312 | ST-0907 line 11 · your entry 44907 | Unexplained, for the AP team |
Only the third line needs a person. The first two are timing and would be noise if they arrived in the same queue.
The human review
An accounts payable specialist works the unexplained differences and decides what to raise with the supplier. Ledger corrections, write-offs and payment releases stay in your finance process.
Set it up for your team
Fix the cut-off rules
Write down how goods in transit, part payments and post-dated credit notes are treated, and the statement date the reconciliation runs against.
Agree the document identifiers
Suppliers number their documents differently from your ledger. Supply the mapping rules and known prefixes rather than leaving the agent to infer them.
Set the output destination
Define the table your team works from and the identifiers it needs to open both original records.
Check the result before scaling
Run a period you have already reconciled by hand and compare the two results. Count the differences found, the differences missed and the explanations that were wrong, before using the output unsupervised.
Keep the data in the right place
Supplier statements expose your purchasing volumes, payment behaviour and commercial terms. A local deployment keeps both records inside your network. A connected ledger or a hosted model needs its own explicit data-access configuration.
Explore deployment optionsQuestions & answers
Will it correct our ledger automatically?
No. It prepares an explained list of differences. Every posting, correction and write-off stays inside your existing finance process.
What if the supplier numbers documents differently from us?
Supply the mapping rules and known prefixes during setup. Lines it cannot align with confidence are reported as unresolved rather than paired on a guess.
Can it read a scanned statement?
Scans can be processed, but accuracy depends on the quality of the original. Validate the extraction on your own suppliers and route unreadable lines to a person.