Small-business outsourcing research

What share of supplier invoices arrive without a purchase order?

A measurement protocol for invoice matching gaps, with channel, supplier type, and the time spent resolving a missing order.

What share of supplier invoices arrive without a purchase order?

Research question and scope

This protocol asks what share of supplier invoices arrive without a purchase order and how much work the gap creates. The unit is one invoice received during the window, identified by the supplier and the invoice number. The study measures the share without a matching order, the reason, and the time to resolve. It does not judge whether the purchase was necessary or whether the supplier is at fault.

Set the scope to one small business, one accounts-payable process, and one supplier base. Record the systems that can receive an invoice and the rule for matching an order. If the approval process changes during the window, keep the earlier definition for the full sample or analyze the change as a separate period. A new order requirement can look like a worsening trend when it is only a different gate.

Define the population as every invoice recorded during the window, including invoices that were rejected, returned, or paid without an order under an exception. The unit is the invoice, not the supplier, because one supplier can send several invoices. A credit note is a separate unit with a link to the original invoice. Test the definition on two recent invoices before collection begins.

Definitions that make the measure reproducible

Define an invoice as received when it enters the business's invoice inbox or accounting system, not when someone opens it. Define a matching purchase order as an approved order that names the same supplier and covers the invoiced items and amount within tolerance. Define resolution as the point when the invoice is matched, approved as an exception, or rejected. Each definition needs a positive and a boundary example.

Do not count a verbal approval as a purchase order unless it was recorded in the system before the invoice arrived. Do not count an order raised after the invoice as a match for the initial state; record it as a late order with the date. A partial match should be recorded with the matched and unmatched amounts. The definitions should survive a second reviewer.

Record the supplier category, the spend band, the invoice channel, and whether the supplier is new or established. These strata matter because a new supplier may not know the order process while an established one may be testing a shortcut. Include invoices that arrived by email, portal, and post. Use one category list for every invoice.

Evidence standard and data dictionary

For each invoice, capture the supplier, the invoice number, the amount, the date received, the order number if any, and the matching status. Add the time to first review, the time to resolution, the resolver, and the reason for the gap. Record whether the invoice was paid under an exception and who approved the exception. Keep the invoice image with the record.

Also capture whether the invoice was a duplicate and whether the supplier sent a statement that included other invoices. These details change the workload even when the gap rate is the same. A second reviewer should be able to rebuild the matching decision from the record and the order system. If the record cannot support that, the sample is incomplete.

If the business uses a purchase-card feed, record whether the card transaction was treated as an order. A card receipt is evidence of a purchase but not necessarily an approved order. The protocol should state which evidence standard applies so the result is comparable across suppliers. Write the standard in the collection guide before the first invoice is sampled.

Sampling and collection

Draw the sample from the invoice log rather than from a list of problem suppliers. Record the export date, the filters, the row count, and the selection rule. Include invoices from every channel and every week so the result is not dominated by a month-end batch. If the population is ordered by supplier, take a fixed interval across the list.

Collect the invoice and the matching record as they existed at the time, not after a later cleanup. If an order was raised after the invoice, record both dates and treat the initial state as unmatched. Do not infer an order from a payment. Note the collection date separately from the invoice date.

A useful additional measure is the share of invoices that were complete on the first submission. A complete invoice names the order, the items, the amount, and the tax. Record the missing fields and the time spent asking the supplier. That view separates a supplier-education problem from an internal approval problem.

Analysis and reporting

Report the number of invoices, the share without a matching order, the share matched late, and the share paid under an exception. Show the counts by supplier category, spend band, and channel. Present the median time to resolution and the share that waited more than five business days. Use the same denominator for every rate.

Inspect the invoices without an order and group the reasons without naming the supplier or employee. A cluster from one channel may point to a portal that does not require an order number. A cluster at month end may point to a rush that bypasses the process. Report duplicates separately because they carry a different control risk.

A useful additional view is the time from the invoice date to the order date when an order was raised later. A long gap suggests the purchase was made without a plan, while a short gap suggests a paperwork delay. That view helps the owner decide whether to fix the buying habit or the order template. Report it beside the overall gap rate rather than in a separate memo.

Limitations and sources of error

The record cannot show whether the purchase was legitimate or whether the price was fair. It can show only whether an order existed and how long the mismatch took to resolve. A high match rate is not proof of good purchasing, because the order may have been raised after the fact. The protocol also cannot capture a purchase made on a personal card and never entered.

Invoice systems can merge suppliers or change an invoice number during import. A duplicate may appear as a new invoice, or a credit may be missed. Record those cases and their treatment instead of dropping them. If the same person raises the order and approves the invoice, note the independence limitation and analyze that group separately.

A useful check is whether a new assistant can find the order for a typical invoice in under two minutes. Ask them to try five invoices and record where they looked. That test measures the system and the naming rule, not the assistant. Fix the search path before asking for more matching discipline.

Review cadence and next action

Give the result to the owner and the accounts-payable lead and ask them to choose one observable change. The change might be a supplier reminder, a required order field on the invoice portal, a pre-approved order for recurring supplies, or a weekly exception review. Record the owner, the effective date, and the measure that will show whether the change worked. A change without a measure becomes an opinion at the next review.

Repeat the protocol with the same definitions after the change and compare the same fields. If the gap rate fell but resolution time did not, the process may still be slow after the order arrives. If resolution time fell but the gap rate rose, the team may be approving exceptions faster without fixing the cause. Close the review when the owner accepts the finding and the next sample is scheduled.

A useful additional view is the share of invoices that were matched by the person who raised the order. A match performed by the buyer can be faster but less independent than a match performed by accounts payable. Record the matcher and the amount. That view helps the owner decide where a second check is worth the time.

Ask the assistant to keep a short list of the order numbers that were hardest to find. The list is a map of the naming and search problems that slow the queue. Review it with the buyer and fix one naming rule at a time. A better index can reduce the gap rate without changing the purchasing policy.

Sources

Frequently asked questions

Does a favorable sample prove the routine caused the result?

No. The design is observational and supports a bounded operating decision, not a universal causal claim.

What should an owner review first?

Start with high-consequence defects, missing denominators, exceptions, and cases that required the owner to reconstruct the evidence.

Related research