Fewer Crashes, Costlier Repairs, and What That Does to a Shop

Short answer: Police-reported crashes fell 8.5 percent between 2019 and 2024 while miles driven rose 1 percent. The repairable claims that actually reach a body shop fell faster: by five to ten percent in each of 2024 and 2025 alone, on the separate estimates the one public filer and the estimating vendor put on the record. Something between the crash and the repair order is removing the work.

By Anthony Braswell for Quorum Industries LLC, The Autobody Directory · Updated 2026-09-01 · How this was written, and what the machine may not do

How this page was produced: Researched, drafted and checked with AI assistance under human direction, and signed off by the named author. How this site is written

Before anyone argues about labour rates, there is a prior question no repairer controls: how many damaged cars get repaired at all. This is also the page where the evidence cuts against the trade's usual argument.

Tiers. [REPORTED] is the source's own statement; [ADJUDICATED] is law.

The crash count is down, and one line of it is up

[REPORTED] NHTSA takes fatal counts from the Fatality Analysis Reporting System, a census, and injury and property-damage-only counts from the Crash Report Sampling System, a sample of police reports.

The arithmetic is mine, from those rows. Total crashes fell 575,843, or 8.5 percent. Property-damage-only crashes, the category closest to what arrives at a body shop, fell 7.1 percent, and injury crashes fell 12.5 percent. Fatal crashes rose 8.4 percent, and any sentence saying crashes are down without that clause misdescribes the record.

[REPORTED] Americans drove further. FHWA Highway Statistics 2023, Table VM-202, gives 3,261,772 million vehicle-miles in 2019; NHTSA puts 2024 at 3,294,031 million and reproduces FHWA's 2023 figure exactly, the cross-check letting them share a column. Miles rose 1.0 percent, and crashes per 100 million vehicle-miles went from 207.13 to 187.62, down 9.4 percent.

Now the caution. [REPORTED] NHTSA's 2024 note reads: "The estimated number of police-reported motor vehicle traffic crashes increased 0.7 percent from 6,138,474 in 2023 to 6,180,241 in 2024, as presented in Table 2." And under its summary table: "Note: Changes from 2023 to 2024 were not statistically significant." The five-year fall survives that. The single-year direction does not. NHTSA also revises between publications, which is why each row above names its own.

What arrives at a shop is not a crash

A crash is not a claim, and a claim is not a repair order. Boyd Group Services Inc. owns its shops, files audited statements and publishes a quantified estimate of repairable claim volume, and that combination is why its filings carry this section. No other collision consolidator located for this page does all three. Boyd is not the trade's only public filer, and this page does not say so: Driven Brands Holdings Inc. files a Form 10-K with a Paint, Collision and Glass segment, and its risk discussion is quoted below.

[REPORTED] From the management discussion filed as Exhibit 99.2 to Boyd's Form 40-F on 2026-03-18: "Based on claims processing platform data, the Company estimates a year-over-year decrease in repairable claims within the range of 5-7%, with the negative claims environment improving throughout the year, ending down within the range of 2-4% in the fourth quarter of 2025." Within the year the path ran 9-10 percent down, then 6-8, then 3-5. By the first quarter of 2026 it was "in the range of 0-2%", and in the second quarter, reported on 2026-08-12, "repairable-claims volumes were flat to down 2% year-over-year".

The 2024 figure in that document is a different kind of statement and is separated here for that reason. [REPORTED] Boyd writes: "This is consistent with market trends where industry sources report a year-over-year decrease in repairable claims within the range of 5-7% (2024 - 7-9%)." That 7 to 9 percent is what Boyd reports industry sources saying, not what Boyd estimates from its own platform data, and it is used below only in that form.

This bound travels with all of those numbers. Boyd does not name the platform, and they are estimates of a quantity the company does not itself measure: not audited counts, not a government statistic.

A second private estimate of the same year exists, from a different company, on a different base. [REPORTED] CCC Intelligent Solutions, Crash Course 2026: "Repairable claim volume declined by 9.7% in 2025 (all coverages), while non-comprehensive volume was -8% for the year." CCC counts estimates written on its own platform; Boyd counts an industry quantity from an unnamed platform. They are not the same measurement and nothing here averages or reconciles them. They agree only on sign and rough order.

Set that beside the crash count. In calendar 2024 NHTSA's crash estimate rose 0.7 percent, a movement it says was not significant, while Boyd filed that industry sources put repairable claims down 7 to 9 percent. We do not claim those measure the same quantity, and nothing here subtracts one from the other. What can be said is the direction: in a year when the government's crash estimate did not fall, the claim count shops live on fell by most of a tenth on the figure Boyd filed.

Boyd names the candidates. [REPORTED] From its Annual Information Form, Exhibit 99.1 to the same 40-F: "The automobile collision repair industry can experience a decrease in repairable claims, higher total loss rates as well as a deferral in repairs and an increase in unfiled claims." And later: "repairable claims volumes have been and may continue to be impacted by an increased number of non-repairable claims or total losses." Read the tense. Three of those channels are consumer decisions, two are insurer decisions, and none is a crash.

The places competing for that work did not shrink. [REPORTED] The BLS Quarterly Census of Employment and Wages, private, national, NAICS 811121, counts 34,904 establishments in 2019 and 35,422 in 2024, employment rising from 245,923 to 257,163. My arithmetic: establishments up 1.5 percent and employment up 4.6 percent across the five years crashes fell 8.5 percent. Crashes per establishment went from 194 to 174.

Total loss is a ratio, and it has swung both ways

A totalled car is a claim paid and a repair not performed: the only mechanism that removes a job after the crash.

There is no government total-loss series. Not NHTSA, not BTS, not FHWA, and none reachable in this research. Every figure the trade quotes comes from a commercial estimating vendor, and the strongest form of one is a figure quoted inside a filed annual report, because the filer is accountable for it.

[REPORTED] IAA, Inc., Form 10-K for fiscal 2022, under "Increases in Vehicle Complexity and Total Loss Frequency": "Based on data from CCC Information Services, the percentage of claims resulting in total losses was approximately 18% in 2022, 20% in 2021 and 21% in 2020."

Twenty-one, twenty, eighteen. Total-loss frequency fell across exactly the period when vehicles were becoming more complex and costlier to fix. Anyone arguing that severity mechanically raises it has to explain that. Total loss is a ratio, and CCC describes what the denominator did in those years. [REPORTED] CCC, Crash Course 2026: "Inflationary conditions in 2021 and 2022 led to unprecedented increases in used vehicle values, which was reflected in the valuations for potential total losses."

Then the denominator turned. [REPORTED] CCC Intelligent Solutions, Crash Course 2025 Q2: "Through April 2025, 22.6% of all losses - and 23.5% of non-comprehensive losses - were declared total losses, a 0.9-point increase year-over-year. Contributing factors include declining used vehicle values, shifting vehicle age mix, and lower claims volumes." And: "The share of claims flagged total loss was a record in 2024 - based on CCC's historical trends."

[REPORTED] The full year closed higher again, and CCC's own qualification is part of the same sentence: "The share of claims flagged total loss increased by 0.8 percentage points to 23.1% across all loss categories, while non-comprehensive claims flagged total loss increased by 1.0 percentage points to 23.9%. This represents a new high watermark for the industry, while being mindful that this ratio is being affected by claim filing behaviors (notably lower-severity 1st party APD claims)." Read that clause twice. A ratio rises when small claims stop being filed, without any car being harder to fix.

We are not claiming those figures form one series. IAA describes the percentage of claims resulting in total losses as at fiscal 2022, attributed to CCC Information Services. CCC's own 2025 report gives shares through April of a partial year. CCC's own 2026 report gives full-year 2025 shares, worded "across all loss categories" and "non-comprehensive". The wordings are not identical, the periods are not the same length, and this page does not combine, difference or plot any of them together. Read separately, each shows a mechanism: when used values rose the share fell, and when they fell it reached a high CCC itself attributes partly to what consumers stopped reporting. CCC's stated causes are vehicle values, vehicle age mix, claim volumes and filing behaviour. They do not include repair complexity. Severity is real. It is not sufficient.

The mechanism is not CCC's alone to state. [REPORTED] Driven Brands Holdings Inc., Form 10-K for the year ended 2025-12-27, listing what its demand depends on: "The used car market, and the average value of used cars, impacts how often cars are deemed a total loss by insurance companies. As the price of used cars decreases, the number of cars being deemed a total loss increases, resulting in less demand for repairs and maintenance." That is a second filer putting the denominator, not the numerator, at the centre of the same decision.

The severity half is stated by IAA in the same 10-K. [REPORTED] IAA's risk factors: "These technological advancements have resulted in higher repair and part replacement costs following an accident, making insurance companies more likely to declare a damaged vehicle a total loss." [REPORTED] And Allstate's Form 10-K for the year ended 2025-12-31 lists what it prices for in vehicle physical damage: "inflation, supply chain disruptions, labor shortages, labor rates, tariffs impacting vehicle and parts prices, increased repair costs for components that have embedded advanced driver assistance systems such as cameras and sensors, length of claim resolution, delays in the receipt of third-party carrier claims, and a higher mix of total losses".

Who actually makes the call

[REPORTED] Copart, Inc., Form 10-K for the fiscal year ended 2025-07-31, Item 1 Business:

"The adjuster determines whether to pay for repairs or to classify the vehicle as a total loss based upon the adjuster's estimate of repair costs, vehicle's salvage value, and the PAV, as well as customer service considerations."

PAV is pre-accident value. Salvage value and pre-accident value are market quantities. The estimate of repair costs is not: it is a document somebody writes, on software, under rules that vary by state, which is the subject of a separate page here. The estimate is the numerator of the ratio deciding whether the car is repaired at all, and the closing clause about customer service means it is not even a formula.

What a repair now has to include

[REPORTED] CCC, Crash Course 2025 Q2: "calibration procedures appearing on over 31% of DRP estimates in Q1 2025, up significantly from 23.9% a year ago". Scans reached "nearly 87%". [REPORTED] CCC, Crash Course 2026, on the wider base of every repairable claim rather than direct-repair estimates: "Across all repairable claims, calibrations saw an almost 6.5 percentage point increase, from 21.8% to 28.3% for the year. This equates to a 30% increase in the number of repairable estimates where a calibration was present." Those are two different denominators and are printed here as two figures, not one trend. They agree on direction. They are not the same measurement and neither is averaged with the other. The 23.9 percent in this paragraph is calibrations on direct-repair estimates a year before Q1 2025; the 23.9 percent in the total-loss section above is a share of non-comprehensive claims. Same digits, unrelated quantities.

[REPORTED] Back in the 2025 Q2 report: "the majority of calibrations appear on supplements, requiring additional review and approval time while vehicles are in the process of being repaired." The fastest-growing line on the estimate is the one argued about after the car is apart. Meanwhile the job shrank. [REPORTED] For the full year 2025 CCC records labour hours "down 0.8 hours per repair in 2025, following a very slight decrease (-0.1 hours) in 2024", and average parts per non-comprehensive repairable appraisal "13.0, down from 13.6 in 2024 (and a peak of 13.7 in 2023)".

The floor under all of this now rises by law. [ADJUDICATED, as a promulgated federal standard] 49 CFR 571.127 "establishes performance requirements for automatic emergency braking (AEB) systems for light vehicles", applying to passenger cars and other light vehicles at or under 4,536 kilograms gross vehicle weight rating. Vehicles built on or after 1 September 2029 must comply, with a year's extension for small-volume manufacturers, final-stage manufacturers and alterers.

Boyd puts a number on the mechanism in its base case. [REPORTED] Boyd's first quarter 2026 results release, Outlook, quoted in full because a growth framework quoted in fragments is a different claim:

"The Company's long-term growth framework contemplates average same-store sales growth of 3-5%, supported by continued incremental market share gains driven by ongoing consolidation within the highly fragmented collision repair industry, Strong performance with insurance clients, and disciplined operational execution. The framework also assumes 3-4% annual growth in average total cost of repair and approximately 1% growth in miles driven, partially offset by an approximate 2% decline in repairable claims due to the impact of collision avoidance systems. While growth in average total cost of repair has remained below historical averages in recent periods, management believes a return toward target levels over time is supported by the continued normalization of key industry drivers, including rising used vehicle values and increasing vehicle complexity."

The capitalised "Strong" mid-sentence is in the filed document and reproduced as printed. Severity up three to four per cent a year, miles up about one, repairable claims down about two with the cause named. A permanent volume decline is not a risk in Boyd's model. It is an assumption in the base case.

That release is the last filing in which Boyd printed the framework. The second quarter 2026 release of 2026-08-12 does not restate it and does not use the words growth framework; it says the quarter's claim volumes were "consistent with our long-term planning assumptions" instead. The framework quoted above is the current published version because nothing has replaced it, not because it was reaffirmed.

The squeeze, stated at its strongest

There is a complete economic explanation of consolidation here needing no villain. Volume per shop is flat to falling while capability per repair rises. A shop that cannot calibrate must sublet, and a shop that sublets gives away the fastest-growing line on the estimate. If fixed capability cost per site rises while jobs per site fall, the arithmetic has one output: fewer, larger, better-capitalised sites.

[REPORTED] The Massachusetts Auto Body Labor Rate Advisory Board's report of 22 December 2025 records, in its section headed "AUTO BODY REPAIR INDUSTRY'S POSITION" and introduced as coming from "another auto body shop representative's written submission": "Small, independent shops are closing or consolidating, unable to reinvest in training, tooling, or environmental compliance required under state and federal law." That is one shop's written submission, reproduced by a state board. It is not a finding by the Board and it is not a survey of anybody. And the consolidator's results look like the squeeze too: [REPORTED] Boyd's net earnings fell from 86,656 thousand US dollars in 2023 to 18,420 thousand in 2025, a fall of 78.7 percent by my arithmetic, while sales rose 6.7 percent.

That is a strong argument. Now test it.

The negative control, and it cuts against the argument above

If capability cost were driving consolidation we would expect a falling gross margin, calibration as a burden, heavy maintenance capital, management blaming capability, acquisitions justified by equipment, regulators writing capability into the definition of a competent shop, and the price of a repair accelerating. Eight places the record looks like the opposite.

One. The gross margin did not deteriorate. [REPORTED] Boyd's annual reports give gross profit of 44.8 percent of sales in 2021, 44.7 in 2022 and 45.5 in 2023; the 40-F states "Gross Profit was $1,458.6 million or 46.4% of sales for the year ended December 31, 2025 compared to $1,396.5 million or 45.5% of sales for the same period in 2024." Not a clean upward march, since 2021 sits above 2022 and 2023 and 2024 are level. But a business crushed by the cost of new capability does not finish a multi-year volume decline at its highest gross margin of the period.

Two. Calibration is a profit centre. [REPORTED] "Gross margin percentage increased due to several factors, including the benefits of internalization of scanning and calibration, increase in parts margin, and improvements in performance based pricing." The same filing sizes it: scanning and calibration "represented 6% of total revenues in 2025". And: "In the fourth quarter of 2025, 75% of Boyd's scanning and calibration services were completed utilizing internal resources in the U.S. business, up from 53% in the fourth quarter of 2024". By the first quarter of 2026 Boyd had "Achieved targeted level of 80% internalization". In the second quarter it said it intends to keep spending on it: growth is to be complemented "together with continued investment in glass, scanning, calibration and other adjacent capabilities".

Three. The consolidator says calibration lowers the insurer's cost. [REPORTED] "Initiatives such as the internalization of scanning and calibration services, progress in Boyd's repair first strategy and focus on the use of cost effective alternative parts, deliver strong value by lowering repair costs for the Company's customers and providing incremental gross margin to Boyd." Boyd's customers are insurers. That is the inverse of the severity story as usually told.

Four. The capital number is small, and it is quoted here with the exclusions it carries. [REPORTED] "The Company's need to maintain its facilities and upgrade or replace equipment to meet increased complexity of newer vehicles, signage, computers, software and vehicles forms part of the annual cash requirements of the business." Then the figure, with the clause the sentence opens on, because without it the number means something else: "Excluding expenditures related to network technology upgrades and acquisition and development, the Company spent approximately $51.7 million, or 1.6% of sales on capital expenditures during 2025, compared to $62.3 million or 2.0% of sales during 2024." The excluded item is disclosed separately: "During 2025, the Company spent approximately $10.7 million on capital network technology upgrades. The investment expected in 2026 is in the range of $2 million to $4 million." Guidance for 2026 is "within the range of 1.6% and 1.8% of sales", again excluding acquisition and development, and again with network technology on top.

My arithmetic, and it weakens the point. Add the network technology line back and Boyd spent about $62.4 million in 2025 outside acquisition and development, which on sales of $3,142,794 thousand is about 2.0 percent, level with 2024 rather than below it. The claim that survives is narrower than the headline figure: everything Boyd spends outside buying and building shops, including the technology and security infrastructure, is about two per cent of sales.

Five. Management blames volume, not capability. [REPORTED] "The decline in net earnings in 2025 compared to 2024 and 2023 was driven by reduced repairable claims volumes which resulted in decreases in both same-store and new location sales." And: "This resulted in decreased leverage in the absorption of fixed costs." Not equipment.

Six. What the acquisitions buy is people. [REPORTED] Boyd's audited financial statements, acquisitions note: "A significant part of the goodwill recorded on the acquisitions can be attributed to the assembled workforce and the operating know-how of key personnel. However, no intangible assets qualified for separate recognition in this respect." If capability equipment were the scarce asset the purchase price allocation would say so, because equipment is exactly what qualifies for separate recognition.

Seven. The one regulator that defined a competent shop did not mention any of it. [ADJUDICATED, as a promulgated state regulation] California's 10 CCR 2695.81 says: "Only labor rates reported by auto body repair shops that meet each of the specific standards set forth in subdivision (d)(4)(A) below may be used in a Standardized Labor Rate Survey." Those eight standards are Bureau of Automotive Repair equipment requirements; garage keeper's liability and workers' compensation cover; "electrical or hydraulic equipment capable of making simultaneous multiple body or structural pulls"; "a spray booth that meets current federal, state, and local requirements"; verified four-wheel alignment; removal and reinstallation of frame, suspension, engine and drive train components; EPA compliant air conditioning service; and subscription to "a provider of structural specifications with periodic updates covering the vehicle structure for the make, model, and year of the vehicle(s) being repaired".

Not one of the eight names calibration, scanning, ADAS, aluminium or carbon fibre. The list is printed so the absence can be checked. "Aluminum repair labor" and "Carbon fiber labor" appear elsewhere, at (d)(8)(A)3, as labour types a shop may quote a rate for, which is not a capability it must have. The absence claim is bounded to subdivision (d)(4)(A), not the whole regulation, in force in this form since 1 January 2017.

Eight. The price of a repair decelerated while the volume fell. [REPORTED] CCC, Crash Course 2026: "Preliminary average total cost of repairs (TCOR) is $4,818 for 2025, an increase of 1.7% over 2024, representing the lowest percentage increase since 2017." Boyd's own framework assumes three to four per cent a year, and Boyd's release concedes the shortfall in the paragraph quoted above. A severity story that cannot show severity rising in the year it is invoked is not a complete story.

Two places the record does support the severity story, which is why this is a control and not a rebuttal. First, three unrelated public filers put the same causal sentence in their own filings: IAA and Allstate as quoted above, and [REPORTED] Copart, in the 10-K already cited here: "We believe that one effect of these additional features is that newer vehicles involved in accidents are more costly to repair and, accordingly, more likely to be deemed a total loss for insurance purposes." Note the verb. Copart says it believes it. Second, Boyd prices the mechanism into its base case with a number attached.

What that leaves standing

The mechanism is documented. Its causal role in consolidation is not. We are not saying rising severity is consolidating the collision repair industry. That is an inference the record does not carry.

What the record carries is narrower. Repairable work is falling faster than crashes are, each repair is changing composition faster than size, total loss is a ratio whose denominator has moved both ways within five years, and the filer that publishes a forecast has booked a permanent volume decline into it and named the cause.

We are not saying any insurer, vendor or consolidator has done anything improper. Every quotation here is a party describing its own business in a document it filed.

What we could not establish, and the walls we hit

No total-loss percentage here is a government statistic. Every one originates with CCC, named in the IAA filing as CCC Information Services and in its own reports as CCC Intelligent Solutions, across two of those reports and one 10-K that quotes it, and they are not joined. Nor could we read the insurer-specific count a federal system does publish. The National Motor Vehicle Title Information System's FY2024 report states that "During FY2024, 16.9 million records were reported by JSI entities", and breaks them out by entity type, including an Insurer category. Those breakouts are charts this instrument could not read. That corrects the research file behind this page, which said the figure was not broken out by insurer. It is; we could not read it.

No federal series measures collision repair cost. Both BLS data hosts, data.bls.gov and api.bls.gov, are robots-disallowed, read before any request; the establishment counts came from www.bls.gov, which permits those paths. fhwa.dot.gov disallows /policyinformation/travel_monitoring/, so Table VM-202 was used instead. federalregister.gov/documents/search is robots-disallowed and its API returns HTTP 403 here, so only the codified braking standard was read and nothing about that rule's economics appears above.

No absence claim about any large SEC filing appears on this page. Every 10-K quotation above is a presence claim. The retrieval tool used to draft this page truncated filings over a few megabytes; the checking pass that followed did not, and each filing cited was downloaded whole, flattened to a single line and string-matched, including the 8 MB Allstate 10-K and the 9 MB Driven Brands 10-K. Both instruments were controlled the same way, with a nonsense string that had to return zero and a known phrase that had to return one, before any count was believed. Even so, nothing here rests on a phrase failing to appear in a filing.

Robots files that would not serve, recorded rather than worked around. https://s25.q4cdn.com/robots.txt returns HTTP 403, so the two Boyd annual report PDFs on that host were fetched without a robots file to read first. https://efts.sec.gov/robots.txt returns HTTP 403, so the checking pass did not use the EDGAR full-text search index at all and matched every filing quotation against the filing itself. https://crashstats.nhtsa.dot.gov/robots.txt returns HTTP 404: no file, so no restriction inferred and none applied. And www.bls.gov served the two establishment-count files to one request and returned HTTP 403 to two others differing only in user agent, which is a reason to record the retrieval and not only the number.

Every filing-based finding is about companies that let themselves be measured. Caliber, Crash Champions and Classic Collision publish no comparable disclosure, and nothing on this page should be read as describing them.

Corrections

These entries correct this page's own draft and the research behind it, before publication.

2026-09-01, en dashes transliterated to hyphens, disclosed rather than done silently. Boyd's first quarter 2026 release prints "3-5%" and "3-4%" in the growth framework paragraph with en dashes, Unicode U+2013, and this site is ASCII only, so both appear above as hyphens. The substitution was checked character by character: that paragraph holds exactly two en dashes and one typographic apostrophe, and the "0-2%" before it already uses an ASCII hyphen in the original. The same transliteration was applied to typographic apostrophes and quotation marks throughout, and to en dashes in the CCC quotations. No other character changed and no word was added or removed.

2026-09-01, two quotations that were wrong in the research behind this page. Copart's adjuster sentence had been cut short at "and the PAV"; the filed sentence continues ", as well as customer service considerations", and two retrieval passes disagreed before a third and two exact-phrase EDGAR searches settled it against accession 0001628280-25-042946. Boyd's internalisation figure had been dated "In 2025" where the filing says "In the fourth quarter of 2025".

2026-09-01, three figures and one gap list corrected. The research file recorded calibrations on "31.7 per cent" of estimates and scans on "86.8 per cent"; a verbatim fetch of the CCC report and a string search find neither, so CCC's own "over 31%" and "nearly 87%" are used. It said Boyd's gross margin "rose every year of the series", which it did not. It carried a per-location capital expenditure illustration on a mismatched base, dropped because Boyd discloses the actual figure. And it listed total-loss frequency after 2022 as unestablished, when CCC gives 2024 as a record share and full-year 2025 as a new high watermark.

2026-09-01, an establishment series re-pulled from a permitted host. The BLS counts were first taken from data.bls.gov, which its own robots.txt disallows in full. They were re-retrieved from www.bls.gov, and both years reproduced identically.

The entries below correct this page's own first published draft, after an adversarial check by a second reader who did not write it.

2026-09-01, a false claim that Boyd is the only public filer. The draft said Boyd was "the only large North American collision consolidator filing audited statements and a written risk discussion". Driven Brands Holdings Inc. files a Form 10-K with a Paint, Collision and Glass segment and a written risk discussion that states the total-loss demand mechanism directly. The sentence has been corrected and the Driven Brands risk factor is now quoted in the total-loss section.

2026-09-01, an estimate attributed to the wrong speaker. The draft reported the 7 to 9 percent fall in 2024 repairable claims as Boyd's. In the filed document that figure is not Boyd's platform estimate but Boyd's report of what "industry sources report", inside a parenthesis in a different sentence from the one carrying Boyd's own 2025 estimate. Both places using it now say so.

2026-09-01, a capital expenditure sentence quoted without the clause it opens on. The draft quoted "the Company spent approximately $51.7 million, or 1.6% of sales" without the words that begin the sentence: "Excluding expenditures related to network technology upgrades and acquisition and development". The full clause is now printed, the separately disclosed $10.7 million of network technology spending is printed with it, and the arithmetic that follows from adding it back, which weakens the point that paragraph was making, is stated in the paragraph.

2026-09-01, a submission described as an industry position. The Massachusetts quotation appears under a section heading reading "AUTO BODY REPAIR INDUSTRY'S POSITION" but is introduced in the report as "another auto body shop representative's written submission". It is now attributed to one submission rather than to the industry.

2026-09-01, four figures were a full reporting cycle out of date. The draft's newest CCC data was the 2025 Q2 report. CCC has since published Crash Course 2026 covering full-year 2025, and Boyd has since reported its second quarter of 2026. The total-loss shares, the calibration share, the labour hours and parts figures and the Boyd claims estimate have all been brought forward, each with its own base named, and the older figures kept where they say something the newer ones do not. CCC's own caution that the total-loss ratio "is being affected by claim filing behaviors" now travels in the same sentence as the record it qualifies.

2026-09-01, an unsourced causal claim sourced. The draft asserted that used vehicle values rose faster than repair cost across 2020 to 2022. That is now carried by CCC's own sentence about 2021 and 2022 rather than asserted by this page.

2026-09-01, a wall statement that was true of one tool and not another. The draft said long SEC filings truncate "in the tools available here". The checking pass downloaded every filing cited in full and string-matched it, including two filings of 8 and 9 megabytes. The paragraph now says which instrument did what. No absence claim about a large filing appeared in the draft and none appears now.

Rejected

Not published here: the 31.7 percent calibration and 86.8 percent scan shares, neither in the CCC report; any figure differencing crash counts against claims estimates; any single total-loss series joining CCC's shares across their different bases, or joining either of them to the percentages IAA quoted; any average of Boyd's and CCC's separate estimates of the fall in repairable claims; any national count of vehicles declared total losses; the braking rulemaking's economics; any federal repair cost index; and any claim about a private consolidator.

Related

• Has consolidation actually reduced the number of body shops
• The shops are profitable, the balance sheets are not
• Total loss: why the first check is low

Sources

• NHTSA DOT HS 813 791, 2024 crash overview. Read on 2026-09-01.
• NHTSA DOT HS 813 435, the 2019 counts. Read on 2026-09-01.
• NHTSA DOT HS 813 560, the restated 2021 figure. Read on 2026-09-01.
• FHWA Table VM-202, 2019 vehicle-miles. Read on 2026-09-01.
• Boyd AIF FY2025, Exhibit 99.1, accession 0001193125-26-112466, the claims risk factor. Read on 2026-09-01.
• Boyd MD&A FY2025, Exhibit 99.2, same accession, claims estimates, margins, internalisation, capital expenditure, earnings attribution. Read on 2026-09-01.
• Boyd financial statements FY2025, Exhibit 99.3, same accession, the goodwill note. Read on 2026-09-01.
• Boyd Q1 2026 release, Exhibit 99.4, accession 0001193125-26-220719, the growth framework, the 0-2 percent estimate and the 80 percent internalisation. Read on 2026-09-01.
• Boyd Q2 2026 release, Exhibit 99.4, filed 2026-08-12, accession 0001193125-26-345758, the flat-to-down-2-percent estimate, the absence of the growth framework and the adjacent-capability investment sentence. Read on 2026-09-01.
• Boyd 2022 Annual Report, 2022 and 2021 gross margin. Read on 2026-09-01.
• Boyd 2023 Annual Report, 2023 gross margin. Read on 2026-09-01.
• Copart Form 10-K, FY ended 31 July 2025, accession 0001628280-25-042946, the adjuster sentence and the more-costly-to-repair sentence, both matched against the whole filing. Read on 2026-09-01.
• IAA Form 10-K for fiscal 2022, accession 0001745041-23-000005, the CCC percentages and complexity sentence. Read on 2026-09-01.
• Allstate Form 10-K, year ended 31 December 2025, accession 0000899051-26-000031, the physical damage list, matched as one contiguous phrase against the whole 8 MB filing. Read on 2026-09-01.
• Driven Brands Holdings Inc., Form 10-K for the year ended 27 December 2025, filed 2026-05-19, accession 0001804745-26-000048, the used-car-value and total-loss demand sentence and the Paint, Collision and Glass segment. Read on 2026-09-01.
• CCC Crash Course 2025 Q2, the through-April-2025 total-loss shares, the DRP calibration and scan shares, the supplement sentence. Its robots.txt permits this path. Read on 2026-09-01.
• CCC Crash Course 2026, Complexity Compounds, full-year 2025: the 23.1 and 23.9 percent total-loss shares with CCC's filing-behaviour caution, calibrations across all repairable claims, the 9.7 percent repairable volume decline, average total cost of repair, labour hours and parts per appraisal, and the 2021 and 2022 used vehicle value sentence. Read on 2026-09-01.
• 49 CFR 571.127, scope, application and 2029 date. Read on 2026-09-01.
• California 10 CCR 2695.81, subdivision (d)(4), the eight standards and (d)(8)(A)3. Read on 2026-09-01.
• BLS QCEW, NAICS 811121, private national annual averages, with the 2019 slice at the same path. Its robots.txt permits these paths. Read on 2026-09-01.
• Massachusetts Auto Body Labor Rate Advisory Board report, 22 December 2025, the shop representative's submission quoted above and the report's own cover date. Read on 2026-09-01.
• NMVTIS 2024 Annual Report, the 16.9 million records and entity-type breakout. Read on 2026-09-01.

Every document above was read on 2026-09-01, and every one of the 55 quotations on this page was matched programmatically against the full text of the document it is drawn from, in a checking pass run by a reader who did not write the page. Where two passes disagreed a third was run, and the disagreement is recorded in Corrections. Anything that would not re-verify was cut rather than softened. Nothing here is legal or investment advice.

General consumer information: not legal, insurance, or financial advice. Requirements, coverage, and practices vary by state, policy, and manufacturer.

More guides