
The Price File and the Contract Disagree
Procurement negotiated the rate. The system holds a different one. The invoice shows a third. Nobody notices, because each individual difference is too small to argue about.
Contract price compliance is the test of whether the price actually invoiced matches the price negotiated in the supplier agreement. It breaks in four places: the contracted price was never loaded into the purchasing system, the price changed and the file was not updated, the buyer ordered outside the contract, or the invoice differs from the purchase order and the tolerance let it through. Negotiated savings that never reach the invoice are the largest quiet loss in most procurement functions.
- Savings are recorded when a contract is signed and realised only when the invoice matches it. The gap between those two events is rarely measured.
- Four break points: price never loaded, price changed without update, buying outside the contract, and invoice tolerance absorbing the difference.
- Tolerance settings are the most under examined control in accounts payable. A small percentage tolerance applied to high volume spend is a large annual figure.
- The test must run at line level against the contract, not at invoice total against the purchase order, because matching a PO only proves you were consistent with your own error.
- ChatFin reads the contract, extracts the agreed pricing, and tests every invoice line against it rather than against the purchase order alone.
Procurement savings have a reporting problem. They are counted at signature, when the negotiated rate is compared to the previous one, and almost never audited at the invoice.
Between those two points sit several handoffs. The agreed price schedule has to reach whoever maintains the purchasing system. It has to be loaded correctly, with the right effective dates, units and currency. It has to be updated when the contract is amended or an indexation clause bites. And the buyer has to actually raise the order against the contract rather than as a free text purchase.
Any of those can fail without a single error message. This page covers the four break points, why matching invoices to purchase orders does not test contract compliance at all, and how to run the test at line level across the whole spend rather than on a sample.

The four places negotiated price leaks
Each break point has a different owner and a different fix, which is why treating price variance as a single problem produces no lasting improvement.
Price never loaded
The contract is signed and the price schedule never reaches the purchasing system, so orders continue at the previous rate. The savings exist only in the report.
Price changed, file not updated
An amendment, an indexation clause or an agreed reduction that nobody applied. The direction of the error is not always in the supplier's favour, which is its own problem.
Buying outside the contract
A free text order or a different supplier for an item covered by an agreement. Common where the contract is not visible at the point of ordering.
Tolerance absorbs the difference
The invoice differs from the purchase order by less than the matching tolerance, so it posts without review. Repeated across high volume spend, this is the largest of the four.
Break three deserves a note. It is frequently misclassified as maverick spend and treated as a behaviour problem, when the underlying cause is usually that the buyer had no practical way to see that a contract existed.
Why PO matching does not test this
Three way matching is a strong control and it answers a different question. It asks whether the invoice agrees with the purchase order and the receipt. It does not ask whether the purchase order was right.
If the wrong price was loaded, the purchase order carries the wrong price, the invoice matches it perfectly and the match passes. The control confirms internal consistency, and internal consistency with an incorrect price is exactly the failure mode being described here.
That last point is the practical obstacle. Testing against the contract means extracting price schedules from agreements held as documents, which historically has meant somebody keying them in, which is why it happens for the top ten suppliers and nobody else.

The tolerance setting nobody revisits
Matching tolerances exist for good reasons: rounding, freight, minor quantity differences, currency conversion. They were set at implementation, usually as a percentage plus an absolute floor, and in most organisations they have never been reviewed since.
| Percentage tolerance | Applied to spend volume, this is an annual figure rather than a per invoice one. Calculate it before deciding it is immaterial. |
| Absolute floor | Often set for low value items and then applied to high volume ones, where it accumulates quickly |
| Direction | Tolerances usually accept differences in both directions, so credits due to you are also absorbed silently |
| Category variation | A tolerance appropriate for freight is rarely appropriate for contracted unit pricing on core spend |
| Review cadence | Almost never revisited after implementation, despite spend volumes and supplier mix changing substantially |
The exercise worth running is simple arithmetic. Take the tolerance percentage, apply it to the annual spend that passes through automated matching, and compare the result to the savings target for the same category. The comparison is usually uncomfortable.
What procurement platforms deliver
Source to pay platforms are strong on contract lifecycle and on catalogue pricing. Coupa, Ivalua, SAP Ariba and Zip manage the agreement, hold negotiated pricing in catalogues, and enforce buying channels that steer users towards contracted suppliers.
Where they work as designed, catalogue buying largely solves breaks one and three. The residual exposure sits in spend that does not flow through a catalogue: services, complex configured items, indexed pricing, tiered rates and anything ordered against a contract in free text. In most mid market organisations that residual is a substantial share of total spend.
Reading contracts and testing invoice lines against them is document work plus matching, which is the combination ChatFin handles, in the same way it clears exceptions described in NetSuite AP automation agents.

Recovery and prevention are different projects
Once overcharges are identified, two separate pieces of work follow and confusing them slows both.
Recovery is commercial. It requires evidence per line, an approach to the supplier, and a decision about which relationships to pursue. Prevention is systems work: correcting the price file, tightening tolerances, and making the contract visible at the point of ordering.
| Evidence pack | Contract clause, agreed price, invoice line, quantity and the calculated difference per line |
| Materiality | Group by supplier before approaching anyone, since one conversation covering a year is better than twelve |
| Relationship | Some differences are worth recording and not pursuing, and that should be a decision rather than an omission |
| Prevention | Correct the price file first, because continuing to overpay while recovering past overpayments is self defeating |
| Tolerance review | Adjust tolerances for contracted categories, which prevents recurrence more effectively than any policy |
The final row usually delivers the largest sustained improvement, and it is the cheapest to implement because it is a configuration change rather than a process change.
How ChatFin tests every line against the contract
ChatFin reads the supplier agreement itself. The contract intelligence and document processing agents extract the price schedule, units, effective dates, indexation clauses and tiered rates into structured records, which is the step that has historically limited this testing to a handful of suppliers.
The autonomous reconciliation agent then tests every invoice line against the contracted price for that item on that date, rather than against the purchase order. Differences are reported per line with the contract reference attached, which is what makes them recoverable. The pattern recognition agent identifies suppliers whose invoiced prices drift consistently in one direction and spend occurring outside any contract at all.
| Contract intelligence | Extracts price schedules, effective dates, indexation and tier structures from agreements |
| Document processing | Reads invoices in any format and structures them to line level with item, quantity and unit price |
| Reconciliation agent | Tests each invoice line against the contracted price rather than against the purchase order |
| Pattern recognition | Flags directional price drift by supplier and spend occurring outside any agreement |
| Governed write back | Prepares the debit note or accrual and the price file correction for a person to approve |
Nothing is sent to the supplier or posted automatically. ChatFin produces the evidence per line and prepares the resulting entry and correction, and a person decides which differences to pursue commercially.

Testing one category properly
Pick a category with high volume and a small number of contracted suppliers. It produces a clear number quickly and the remediation is contained.
Extract the contracted price schedules
Structure the agreed prices with effective dates, units and any tier or indexation terms. This is the step that determines whether the test is possible at all.
Test twelve months at line level
Compare every invoice line to the contracted price for that date. Do not sample and do not net differences in your favour against those against you.
Classify each difference
Price never loaded, price changed, bought outside contract, or within tolerance. Each points at a different fix.
Fix the file before pursuing recovery
Correct the pricing in the purchasing system first. Recovering last year while continuing to overpay this year is not a result.
Teams running this for the first time are usually most surprised by the within tolerance category, because those differences were reviewed by a control that was working exactly as configured.
Three way matching proves the invoice agrees with your purchase order. If the wrong price was loaded, all it proves is that you were consistent with your own error.
ChatFin
Frequently Asked Questions
What is contract price compliance?
It is the test of whether the price actually invoiced matches the price negotiated in the supplier agreement. It differs from three way matching, which compares the invoice to the purchase order and the receipt. If an incorrect price was loaded into the purchasing system, the invoice will match the purchase order perfectly and the match will pass, while the contracted price was never applied at all.
Why do negotiated savings not reach the invoice?
Four common reasons. The agreed price schedule was never loaded into the purchasing system. The price changed through an amendment or indexation clause and the file was not updated. The buyer ordered outside the contract, often because the agreement was not visible at the point of ordering. Or the invoice differed from the purchase order by less than the matching tolerance and posted without review.
How should matching tolerances be set?
By category and with the annual volume in mind rather than the individual invoice. A percentage tolerance appropriate for freight variation applied to high volume contracted unit pricing produces a substantial annual figure. It is worth calculating the tolerance percentage against annual matched spend and comparing that number to the savings target for the same category before concluding the setting is immaterial.
Can ChatFin read supplier contracts to check pricing?
Yes. ChatFin extracts price schedules, effective dates, indexation clauses and tier structures from supplier agreements held as documents, then tests every invoice line against the contracted price for that item on that date. Differences are reported per line with the contract reference attached, and the resulting debit note or price file correction is prepared for a person to approve.
See ChatFin check a year of invoice lines against your agreements.
ChatFin extracts the price schedule from your supplier contracts, including tiers, effective dates and indexation, and tests every invoice line against the price that should have applied.
Differences are reported per line with the contract reference, and the debit note or price file correction is prepared for a person to approve before anything is sent or posted.
Book a DemoBring a contract and twelve months of invoices.
ChatFin will extract the agreed price schedule and test every invoice line against it, classifying each difference by cause and quantifying the recoverable amount.