Blog · Procurement and supply chain · Procurement
Keep missing contract evidence separate from confirmed unmanaged purchases. Report evidence coverage and managed-share bounds on a known spend perimeter.
An invoice that does not match the supplied contract file may be unmanaged. It may also refer to a contract missing from that extract, a supplier alias or a category mapping the file cannot resolve. Treating all three as proven unmanaged buying produces a precise figure with an uncertain basis.
The existing SUM guide owns the selected measure. This page handles incomplete evidence under that selected policy. It does not introduce a new universal definition of managed spend or assume that every contract register is complete.
Confirmed managed means the available, reviewed evidence establishes the selected policy's requirements. Confirmed unmanaged means adequate evidence establishes that the purchase does not meet those requirements. Unknown means the evidence cannot resolve the classification.
Distinguish unknown status from an excluded amount. Excluded spend is outside the stated denominator by policy. Unknown status is inside a known denominator but has an unresolved numerator contribution. If the denominator itself is uncertain, report that uncertainty separately rather than reusing the simple bounds below.
A failed join is a technical observation. It becomes a business classification only after the join keys, contract-file completeness and policy scope are assessed. The zero/multiple contract-match guide describes that investigation.
Useful reasons include missing contract export, unresolved supplier identity, absent category, missing effective dates, unclear unit coverage and multiple applicable records. Keep the reason, amount, owner, requested evidence, review date and current disposition on each line.
Do not convert Unknown to Unmanaged merely to make a chart add up. It already belongs in its own reconciling bucket. Equally, do not label it Managed because the supplier is familiar or a contract exists somewhere in the organization.
Microsoft documents join kinds that retain unmatched rows. Deleting an unknown from the report is an analysis decision; it should require an explicit perimeter rule. Merge queries overview.
The sample contains $100,000 of positive, addressable invoice spend. The denominator is fully observed and fixed; there are no credit notes or uncertain exclusions in this example.
| Evidence state | Amount | Share of known perimeter |
|---|---|---|
| Confirmed managed | $60,000 | 60% |
| Confirmed unmanaged | $25,000 | 25% |
| Unknown | $15,000 | 15% |
| Total | $100,000 | 100% |
Evidence coverage is ($60,000 + $25,000)/$100,000 = 85%. The confirmed managed share is 60%. If all unknown spend proved unmanaged, final managed share would be 60%; if all proved managed, it would be 75%. Report a 60%-75% bound alongside the selected measure and its coverage.
A second figure, $60,000/$85,000 = 70.6%, is managed share among resolved evidence only. It can be useful diagnostically, but it must be labeled as such. Presenting it as the full-book figure removes the difficult 15% from the denominator.
AP later supplies missing records and a reviewed crosswalk. Of the unknown $15,000, $10,000 proves managed and $5,000 unmanaged. The final states are $70,000 managed and $30,000 unmanaged, with 100% evidence coverage.
The managed result is now 70%, but no invoice changed and no new purchase was routed differently. The ten-point rise from the originally confirmed 60% reflects resolved evidence. That is a useful data improvement, distinct from a behavioral procurement improvement.
Preserve the originally issued result, corrected result and reason for change. The reclassification bridge separates evidence, definition and perimeter changes from like-for-like purchasing movement.
The simple lower/upper calculation assumes positive spend with a known denominator. An unallocated negative credit can reduce the numerator, the denominator or both depending on its evidence and policy. Do not add absolute credit amounts to an uncertainty bound mechanically.
Use the finance-approved basis in the credit/tax/currency guide, separate unresolved positive purchases from unresolved negative adjustments, and calculate scenario endpoints under consistent classifications. If a proposed exclusion is unresolved, state both the denominator scenarios and resulting rates.
Prioritize unknowns by value and decision impact. An unresolved $200,000 supplier could materially change category concentration, while a missing code on a $20 invoice may not. Give each case a request that can be answered: provide the dated contract row, confirm this supplier alias, or clarify whether Unit B is covered.
A pilot should test unknown handling rather than hide it; see the procurement acceptance specification. Bring an authorized AP sample and known evidence gaps to discuss the scope. This method supports an honest review, not a promise of automatic contract discovery or complete source coverage.
Not when the file, supplier mapping or scope evidence is incomplete. A failed join remains unknown until sufficient evidence resolves it.
Not mechanically. Separate signed adjustments and apply consistent classification and denominator scenarios under the approved amount policy.