Every company that sells freight audit, including this one, quotes a statistic about how many freight invoices contain errors. One in five. Up to 15%. About 25%. Around 80% of carrier invoices. The numbers appear on vendor sites, in trade articles, in RFP decks and board slides, usually attached to a source name that sounds authoritative.
We spent a research cycle trying to open those sources. The modern freight invoice error rate has not been measured by anyone whose work we could find.
Disclosure, up front
CheckMyFreightBill sells freight audit services. A high published error rate is good for our business. We have the same incentive as every publisher named below to repeat a large number and move on, and we are choosing not to.
That conflict set the rules. We are not replacing the numbers we could not verify with a number of our own, and we are not claiming the error rate is low. We do not know what it is, and neither does anyone who has published a figure we could trace. Unmeasured and low are different claims.
How sources were graded
Each figure was graded on where it came from, not on whether it sounded plausible.
| Grade | Definition |
|---|---|
| A | Government or academic, methodology published, independently verifiable |
| B | Industry body or trade press with disclosed methodology |
| C | Vendor-published with some methodology |
| D | Vendor marketing with no methodology, or an uncitable circular reference |
The tally: seven A-grade sources, three B, three C, and more than twenty D. Every remaining invoice error rate, recovery-as-percent-of-spend figure and accessorial-as-percent-of-spend figure landed in D.
What matters is what the A-grade sources are about. All seven concern detention, demurrage, or 1980s federal overcharge recovery. Not one describes a modern freight invoice error rate.
Claim one: “1 in 5 freight invoices contain errors”
This is the most widely circulated figure in the category, and the one claim with a traceable origin. That origin appears to be the Journal of Commerce’s inaugural Freight Payment Benchmark Study, “Driving Strategic Value Through Outsourcing”, sponsored by IHS Markit and reported by Eric Johnson on 5 October 2020. Shipper respondents reported an average invoice inaccuracy rate of 20.8%.
The methodology is partially disclosed. The study aggregates the responses of 77 shippers across all modes, nearly three-quarters of which generate $1 billion or more in annual revenue, surveyed in summer 2020.
Three things follow, and all three get lost when the figure is repeated. It is self-reported: shippers were asked what they thought their rate was, and no invoices were audited. It never defines an inaccurate invoice, so 20.8% is an average of 77 private definitions. And the sample is self-selected large enterprises, skewing toward complex, high-accessorial programs, surveyed during 2020 COVID disruption. No later edition of this study was confirmed to exist.
The laundering step is documented. A November 2021 LogiSYM article, “Is your freight invoice accurate? A new study claims it might not be”, states that as much as 20% of freight invoices may be inaccurate and that 1 in 5 is significant, attributing it only to “a 2021 Cargo Owners and freight forwarders survey” with no publisher, sample size or methodology. A sourced 20.8% self-report became an unsourced “20% of invoices are wrong”, with the year changed and the sample gone.
The defensible statement is: 77 shippers surveyed by the Journal of Commerce in 2020 self-reported an average invoice inaccuracy rate of 20.8%. “20% of freight invoices contain errors” is a different sentence and is not supported.
Claim two: four named sources we could not find
Four widely quoted statistics carry the name of a real, checkable institution. In each case we went looking for the underlying document and did not find it. A failure to locate is not proof of non-existence, but it is what a buyer checking the claim would experience.
“Up to 15% of freight bills contain errors, and nearly 80% of companies do not have a systematic audit process”, attributed to the Journal of Business Logistics. JBL is a real peer-reviewed Wiley journal. No paper matching this claim was located, and none of the pages citing it give a volume, issue, year, author or title. It is a citation to a journal name, not to a paper, and multiple attempts to trace it failed. The publisher circulating it, Hatfield and Associates (31 March 2025), sells freight audit and payment services and separately claims 15-30% average cost savings for itself. This is the highest-risk claim in the set: attaching a real academic journal’s name to an unlocatable finding is precisely what a skeptical buyer will check.
“22% of freight invoices contain errors requiring manual correction, at an average of $53.50 per error to resolve”, attributed to IOFM, 2025. IOFM’s benchmarking data is member-gated and its site returned an HTTP 403, so this could not be checked at source. Two things make it suspicious regardless. IOFM is an accounts payable body, and freight-specific invoice error benchmarking is outside its normal scope. And a round percentage paired with an implausibly precise dollar figure is a recognizable signature of generated statistics. Every page carrying it is a 2025 or 2026 SEO blog. Unverified, and we would not use it until someone produces the report page.
“5-10% of all invoices contain at least one discrepancy” and “automated systems detect 85-95% of discrepancies versus 40-55% manual”, attributed to an American Shipper 2024 Freight Audit and Payment Benchmark Study. No such study was located. American Shipper is a real publication, and JOC and American Shipper did run the 2020 study described above, which is presumably where the plausibility comes from. No 2024 edition carrying these figures was found; they appear only in recent SEO blogs.
“1% to 5% of all freight invoices contain errors”, attributed to Supply Chain Dive. Supply Chain Dive is legitimate trade press. No editorial article containing this figure was located. Searches of the site surfaced only sponsored vendor content on freight audit topics, on sponsored URL paths. Cited by Hatfield and Associates, 31 March 2025.
Claim three: the NASSTRAC orphan
“Up to 10% of freight bills contain errors, per NASSTRAC” is attributed to the National Shippers Strategic Transportation Council, a real industry body. NASSTRAC’s own site does not publish the figure. No original was found.
Claim four: the circular recovery rate
The claim that a freight audit recovers 2 to 5 percent of freight spend is the number most often used to justify buying the service. Its citation chain is short and closed.
Trax Technologies, in “A Primer on Freight Audit Recovery” dated 20 July 2026, states that between 3 and 6 percent of invoices contain errors, translating to 1 to 5 percent of total freight spend, and hyperlinks Transportation Insight as its source. Transportation Insight is a competing freight audit vendor and publishes no methodology for the figure. Vendor A cites vendor B, vendor B cites nobody, and both sell the service the number justifies.
Adjacent recovery claims found, none with methodology, all from parties selling the service: 3-7% of total freight spend for mid-market shippers; 8-12% of audited spend at 150%-plus Year 1 ROI; 2-8% of audited spend; 1-5% savings; 15-30% average cost savings.
There is no publishable industry recovery rate. Any recovery number on any vendor’s site, ours included, should be that vendor’s own measured result from its own customers, with its own sample disclosed.
Claim five: accessorials as a share of spend
No primary study, trade association benchmark, peer-reviewed paper or government source publishes accessorials as a share of freight spend. Every figure in circulation traces to an uncited vendor blog.
| Figure | Publisher | Publisher’s business |
|---|---|---|
| “Industry average is 22 to 35%” of freight cost | Warp, Freight Spend Analysis Guide | Brokerage / 3PL |
| 12-18% of domestic US freight spend, 8-15% international | DocShipper glossary | Freight forwarder |
| “20-30% of total freight spend for a typical shipper” | FreightPlus blog | Freight management |
| 20-30% of total parcel spend, up to 40% in peak | attributed to SmartKargo data via IndexBox | Software vendor |
The ranges are mutually inconsistent across overlapping scopes. None defines “accessorial”. None states a sample, period or mode mix. Every publisher benefits commercially from the number appearing large. The only named formal study in this space, Mastio & Company’s paid LTL Carrier Benchmarking Study, publishes no percent-of-spend figure at all.
The spread is the evidence
Set the invoice error claims side by side and the range runs from 1% to 80%.
| Figure | Publisher |
|---|---|
| “About 25% of all freight invoices contain errors” | Shipware |
| “between 3 and 6 percent of invoices contain errors” | Transportation Insight |
| “Up to 20% of freight invoices contain errors” | STB Freight |
| 5-8% in manual programs, 1-2% under a TMS | Nuvocargo |
| “error rates exceeding 12 percent of total freight spend” | Trax Technologies |
| “around 80% of carrier invoices contain some kind of discrepancy” | various |
Note that Trax’s figure is a different unit entirely. Every one of those publishers sells freight audit or logistics services. A spread that wide across publishers with an aligned commercial interest is itself the strongest available evidence that no measured population figure exists. If one did, the numbers would cluster.
One data point on the other side: FreightWaves published “Freight Invoice Audits and Why They Matter” on 12 February 2025 as editorial, not sponsored content. It discusses overcharges, duplicate invoices, misclassified freight and unauthorized fees conceptually, and cites no statistics at all. The leading trade outlet declines to put a number on it.
What does hold up
The A-grade sources are real, rigorous and mostly about something else. The gap between what they measure and what they get quoted for is the second finding of this study.
DOT Office of Inspector General, ST2018019, 31 January 2018. Estimates that detention reduces annual earnings by $1.1 billion to $1.3 billion for for-hire truckload drivers, $1,281 to $1,534 per driver per year, and cuts carrier net income by $250.6 million to $302.9 million annually. Methodology fully published, built on 2013 dwell data and FMCSA crash records covering 861,340 carriers and 5,863,573 monthly observations. Two lines get dropped in circulation: “Accurate industrywide data on driver detention do not currently exist”, and “We did not examine the effects of detention on shippers or receivers.” The claim that detention costs shippers over a billion dollars a year is this figure, misattributed. It is a driver earnings estimate.
FMCSA / Virginia Tech Transportation Institute, FMCSA-RRR-13-060, December 2014. Drivers were detained on roughly 1 in every 10 stops, averaging 1.4 hours beyond the 2-hour loading window, and on 19 percent of stops at medium-sized carriers versus 9 percent at large ones. Built from 3,516,816 raw stops filtered to 1,348,897 across 31 carriers, sample composition disclosed. The gold standard in the field: federal, large sample, explicit definition of the measured quantity, published limitations. Nothing on the billing error side is close.
GAO-11-198, January 2011. Of over 300 drivers interviewed, 59 percent reported detention within the past two weeks and about 80 percent of those said it affected their ability to meet hours-of-service limits. GAO characterizes the sample as nongeneralizable.
GAO AFMD-89-47 and NSIAD-88-174. From October 1986 through August 1988, GSA collected $100.8 million in transportation overcharges. A 1988 GAO review found rail carriers overcharged the Army National Training Center by over $3 million: $1.7 million from improper application of negotiated rates, about $1.3 million from improperly prepared government bills of lading. Real recoveries from real audits with causes broken out, which is what does not exist in the private sector. Also pre-EDI and pre-TMS. Historical colour only.
Federal Maritime Commission quarterly detention and demurrage data. Nine ocean carriers report quarterly; the FMC states they collected roughly $15.4 billion in D&D charges between 1 April 2020 and 31 March 2025. The only authoritative dollar figures in ocean D&D. FMC Fact Finding 28, often cited for dollar figures, contains none; it is qualitative.
ATRI’s detention reports (Grade B/C). Methodology published, but self-selected trade-show convenience samples. ATRI’s 2024 report puts detention at 39.3 percent of all stops in 2023 against VTTI’s roughly 10 percent in 2014. That is a definition change, not a trend: ATRI measures waiting beyond the scheduled appointment, VTTI beyond two hours. Never a time series.
APQC, via CFO.com, 3 March 2020 (Grade B). Duplicate or erroneous payments run 0.8% of annual disbursements at top performers and about 2% at bottom performers. APQC is a non-profit with published metric definitions and no stake in freight audit. It is also all-AP, not freight, so applying it to freight is an inference, not a finding.
Independent corroboration
We are not the only party to have run this check and come up empty. GingerControl, itself a freight audit vendor, published a guide in July 2026 stating: “Chase any of them to a primary source and the trail goes cold. NASSTRAC’s site does not publish the 10 percent figure; no dated original backs the 15 percent or the ‘80 percent of carrier invoices’ numbers. They are orphan attributions that vendor blogs quote from each other.”
A competitor reached the same conclusion independently. The same publisher elsewhere cites a 2-8% recoverable-from-audited-spend figure with no external source, which shows how strong the pull toward a quotable number is.
Limits of this work
The search budget was exhausted before every thread closed. Two claims stayed open. The first is a dated, internally consistent set of accessorial figures: that accessorials were an additional 3% on top of linehaul in 2014 and have grown roughly five-fold to 14%, and that about 10% of invoices carried an accessorial charge in 2014 rising to nearly 40%. It reads like real carrier-data analysis, possibly an LTL benchmarking presentation, but its origin was not located and it is unverified. The second is that accessorials represent 20-30% of total parcel spend rising to 40% in peak; no origin was located at all.
Other limits. Ardent Partners’ widely quoted ~14% invoice exception rate could not be confirmed against the primary PDF, so it remains secondhand, and “exception rate” is not “error rate” anyway. APQC’s 1.5% median duplicate-payment figure appears in secondary reporting but not in the CFO.com article, which gives only the 0.8% and 2% endpoints. GAO’s full PDF for GAO-11-198 returned a 403 to automated fetching, so a 65 percent figure ATRI attributes to it is unverified against GAO’s own text. And “not located” is a statement about a search, not proof a document does not exist. If anyone can produce the Journal of Business Logistics paper, the IOFM report page or the American Shipper 2024 study, we will correct this study and say so.
What a shipper should do instead
The honest conclusion is that nobody has published a credible measurement of how often freight invoices are wrong, and the numbers filling that vacuum were written by parties who profit from the answer.
The way out is first-party measurement. Take a defined slice of your own invoices: one quarter, one mode, one carrier group. Write down before you start what counts as an error, such as duplicate billing, rate applied off-contract, fuel surcharge computed from the wrong week or region, accessorial billed without an authorizing event, weight or class differing from the bill of lading. Audit all of them, not a hand-picked sample, and record the count, the dollar value, the denominator and the period.
What you get is a number with a stated method and sample size, describing your own freight, carriers and contracts. That beats every figure in circulation, including the ones in this study, because it is the only one whose subject is you.
Some of those checks are fully deterministic. A fuel surcharge is indexed to a published Energy Information Administration weekly diesel number: either the carrier used the right week and region or it did not. The FMC’s OSRA billing rule and the Surface Transportation Board’s minimum-invoice-information rule for Class I demurrage prescribe what a compliant invoice must contain: either the fields are there or they are not. Those need no industry error rate to justify them, which is convenient, because there isn’t one.