What's a reasonable MRO exception rate?

Executive Summary

There is no published, credible exception rate for MRO and Class C consumables invoices, industry-wide or otherwise. A number offered as a benchmark for how many line items should flag is not measuring your catalog, your vendor mix, or your price tolerance, so it cannot tell you whether your own AP process is working. The mechanism that produces a defensible rate is a comparison: invoiced unit price against the price on your last approved purchase order or contract catalog, line by line, with a tolerance band you set.

Two forces move that band even when your process is unchanged. Vendors substitute parts under a similar description, which resets the price without resetting your reference point. And commodity input costs move: the US Bureau of Labor Statistics Producer Price Index for general purpose machinery and equipment stood at 379.724 in July 2026, up 5.6% year over year (BLS PPI series WPU114, read 2026-09-06), the kind of movement that pushes routine repricing through a catalog even when nothing else about the vendor relationship has changed.

What changes the outcome is not a target percentage borrowed from somewhere else. It is whether the reference price your AP team checks against is current, and whether a substituted part number is caught before it is waved through as a like-for-like repricing.

1. What does an exception rate actually measure on an MRO invoice?

An exception rate is the share of invoice lines that fail a check against a reference price, usually the last approved purchase order or the contract catalog. It measures the gap between the two, not the size of any single overcharge. A high rate can mean your catalog is stale rather than that your vendor is overbilling, and a low rate can mean your check is too loose to catch a substituted part, so the rate alone tells you nothing.

The number itself is a ratio: lines flagged divided by lines reviewed, over a stated period. Nothing about that ratio says whether it is good or bad in isolation. A rate of any size only becomes meaningful once you know what reference price it was measured against and how current that reference was on the day the invoice was checked.

This is why an external benchmark for this category does not transfer. Two companies buying the same fastener from the same distributor can run entirely different rates depending on how recently their catalog was refreshed, how many SKUs they buy against a locked price versus a floating one, and how tightly their AP team enforces a description match on the part number. None of that is comparable across companies, which is exactly why no dataset exists that could responsibly publish one.

2. How do you build a reference price for a consumables catalog?

Start from the last document both sides agreed to: a signed contract price sheet, a purchase order, or a punch-out catalog snapshot with a date attached. Store the unit price against the vendor's part number, not your internal item code, since that is the field the invoice actually carries. Refresh the reference whenever the vendor issues a new price sheet, and log the refresh date so a stale comparison is visible rather than silent.

A reference price is only as good as its source document. A verbal quote or an old email is not enough to check an invoice against with any confidence, because there is no line-by-line price to compare. The contract catalog, the punch-out file, or the most recent purchase order confirmation are the documents that actually carry a unit price per part number.

Keep the vendor's own part number as the match key. Internal item codes get remapped over time and across ERP migrations, and a mismatch there produces a false exception that has nothing to do with pricing. The date the reference was captured matters as much as the price itself: a reference six months stale against a vendor that reprices quarterly will flag legitimate increases as exceptions, which erodes trust in the whole check.

3. Why does substitution pricing break a fixed exception threshold?

A vendor can ship a different part under a description close enough to the original that three-way matching against the purchase order still clears. The price attached to the substitute part was never checked against your reference at all, because the reference was keyed to the original part number. A fixed exception threshold assumes the part being priced is the part you ordered, and substitution is exactly the case where that assumption fails.

Three-way matching checks the invoice against the purchase order and the receipt. It confirms quantity and a general description match. It does not test whether the part number on the invoice is the part number priced on the reference catalog, which is a separate check that has to be built deliberately rather than assumed inside the standard match.

When a substitute is billed at a higher unit price than the original, no threshold catches it unless the check compares the specific part number, not the item category. This is a distinct mechanism from a price increase on the same part, and it needs a distinct control: a part-number match step before the price comparison runs, not folded into it.

• Description drift: The invoice description reads close enough to the ordered item that a manual reviewer approves it without checking the part number field.
• Catalog code mismatch: The vendor's internal SKU for the substitute differs from the one on your reference sheet, so an automated match either fails silently or is skipped.
• Equivalent-grade billing: A lower-spec or generic-equivalent part is billed at the original branded part's price, which a quantity and category match will not detect.

4. How does input cost movement change what counts as a legitimate repricing?

General purpose machinery and equipment input costs measured by the US Bureau of Labor Statistics rose 5.6% year over year to an index value of 379.724 in July 2026 (BLS PPI series WPU114, read 2026-09-06). A reference price set before that movement will read every subsequent invoice as an exception, even where the vendor's increase is proportionate. The fix is not a wider tolerance band; it is refreshing the reference on a schedule that keeps pace with the.

A price increase that tracks input cost movement is not the same finding as an unearned markup, and a static reference price cannot tell them apart. If your catalog was locked six months before a PPI move of this size, every invoice that follows it will look like an exception whether or not the vendor's increase was proportionate to their own cost base.

The practical response is to treat the reference price as something with a shelf life tied to the category's volatility, not a one-time setup task. A consumables category tracking a commodity index that moves several percent a year needs its catalog refreshed on a matching cadence, or the exception check starts flagging normal cost pass-through as if it were drift, and the team stops trusting the flags.

5. Should the exception threshold be a dollar amount or a percentage?

Both, applied together rather than either alone. A percentage threshold catches proportionate overbilling on high-value lines but lets small-dollar consumables drift indefinitely since a large percentage move on a low-price item is still a trivial dollar amount. A dollar floor combined with a percentage ceiling catches both cases: review anything over a set percentage deviation, and separately review anything over a set dollar deviation regardless of percentage.

A consumables catalog spans parts worth a few cents and parts worth several hundred dollars in the same purchase order. A percentage-only rule treats a two-cent increase on a fastener the same way it treats a two-cent-on-the-dollar increase on a bearing, which is not the same finding in dollar terms even though the percentage move is identical.

Running a dual test, a percentage deviation and a separate dollar deviation, catches both failure modes without inflating the exception count on low-value lines that will never matter financially. Which specific percentage and dollar figures to set is a decision only your own spend data can inform, since it depends on your catalog's price distribution and how much reviewer time you can commit to clearing flags.

6. What should you do instead of adopting a published exception rate?

Build the reference catalog from your own contract and purchase order data, key it to vendor part numbers, refresh it on a cadence matched to the category's cost movement, and set a dual dollar-and-percentage threshold sized to your own reviewer capacity. Track the rate your own check produces over several cycles as your baseline, then investigate movement away from that baseline rather than comparing the raw number to any external figure, since no such comparison is meaningful.

The output of this process is a rate specific to your own vendor mix and catalog discipline, which is the only kind of exception rate that tells you anything. Compare this cycle's rate to your own prior cycles, not to a number pulled from outside your data.

A diagnostic that reviews your MRO and Class C consumables invoices against your own contract terms builds exactly this reference catalog as part of the engagement, which is a faster starting point than assembling one from scratch. See the sub-hub on MRO and Class C spend audit for how that review is scoped, and how do you audit MRO and Class C consumables invoices for the line-level checks it runs.

For the wider pattern this sits inside, start with the margin drift guide. See also the six categories drift hides in and accessorial charge audit: the surcharges nobody validates.

7. Frequently Asked Questions (People Also Ask)

Is there an industry-standard exception rate for MRO invoices?

No. Proof.md records no dataset on the distribution of findings by category or vendor type, and no such figure is published anywhere the engine can source it. Any rate quoted as an industry standard for this category is not verifiable and should not be adopted as a target.

What is a reasonable starting tolerance band for a consumables price check?

There is no sourced starting figure to give here. Set the band from your own price distribution: review the range of unit price changes across a recent cycle of invoices against your reference catalog, and set the band where it separates routine repricing from the rest, using your own data rather than an external number.

Does three-way matching catch substitution pricing?

Three-way matching checks the invoice against the purchase order and receipt for quantity and general description. It does not test whether the part number invoiced is the part number priced on your reference catalog, so a substituted part billed at a different price passes the standard match unless a separate part-number check is added.

How often should an MRO price catalog be refreshed?

On a cadence matched to how fast the category's input costs move. The US BLS Producer Price Index for general purpose machinery and equipment rose 5.6% year over year to 379.724 in July 2026 (series WPU114, read 2026-09-06), movement that pushes a static reference out of date within months, not years, for categories tracking that index.

Should small-dollar consumables lines get the same review as high-value parts?

A percentage-only threshold under-reviews small-dollar lines, since a large percentage move on a low-price item is still a trivial dollar amount. A dollar floor applied alongside the percentage threshold catches high-value lines a percentage rule would miss and low-value lines a percentage rule over-flags.

Can a vendor's price increase be legitimate even if it triggers an exception flag?

Yes. An exception flag means the invoiced price differs from your reference, not that the difference is unearned. A price movement that tracks input cost data, such as the BLS PPI series for the category, can be a proportionate repricing rather than an overcharge; the flag starts a review, it does not conclude one.

What document should be used as the reference price for a consumables SKU?

The most recent document both sides agreed to: a signed contract price sheet, a purchase order confirmation, or a punch-out catalog snapshot, each with a capture date attached. A verbal quote or an informal email does not carry a defensible per-part-number price to check against.

What is the difference between a repricing and margin drift on a consumables invoice?

A repricing is a price change the vendor is contractually entitled to make, documented and traceable to a new price sheet. Margin drift is the gap between what the contract or catalog says and what the invoice actually charges, whether from a stale reference, a substituted part, or an unauthorized increase.

Does a diagnostic set a target exception rate for a client?

No target rate is set in advance, since no defensible external figure exists to set it against. A diagnostic builds the reference catalog from the client's own contract and purchase order data and reports what that comparison actually finds, category by category.