Auction Vehicle Repair Cost Estimation

Parent: Used Car Auctions and Buying · Published reference · snapshot 2026-09-19 · skill used-car-buying/references/auction-vehicle-repair-cost-estimation.md

↓ Facts as markdown all context files

Depth-first rabbithole dossier for Auction Vehicle Repair Cost Estimation; source-anchored research pack.

These notes link each claim to its source. A source may be a research report hosted on this site rather than the primary document. A published reference means the content is available; it does not certify independent review or accuracy.Read the editorial policy and follow the sources before relying on a claim.

Definitions

**Claim 3a — AutoGrade is a proprietary, patented Manheim algorithm licensed industry-wide, built on the NAAA grading scale.** The same document states AutoGrade "uses a Manheim-developed algorithm, known as the AutoGrade® Condition Score, and makes this score available to the entire wholesale vehicle inspection industry (U.S. Patent No. 8,230,362)," is "based on the concept introduced in the NAAA Vehicle Condition Grading Scale Guidelines," and is updated on roughly a quarterly cadence. It was initially offered only to NAAA member auctions and later opened to non-NAAA auctions, marketplaces, [source]
**Claim 6 — The 2021–2022 period is the inflection point where computer vision entered auction damage capture, moving the industry toward automated damage detection.** Cox Automotive acquired Fyusion on January 5, 2021; the two had collaborated since 2018 on the imaging in the Manheim Express app, and the press release frames the deal as moving clients "toward automated damage detection," including Manheim's first AI-driven fixed imaging gantry at Manheim Minneapolis. [source]

Structure and components

20. Copart disclaims the figure's accuracy explicitly, and names the mechanism of divergence. The stated caveats are that the figure "may vary greatly from the actual cost to repair this vehicle," because "the availability and use of recycled parts, varying labour rates, paint times, etc. greatly affect the cost to repair any vehicle," that it "is presented for informational purposes only," and that "Copart makes no representations as to the validity or accuracy of the listed value." (https://support.copart.co.uk/faq/what-do-copart-terms-mean/; https://carproxy.com/how-it-works/common-copart-t [source]
1. **1946–1980s — the cost substrate is built outside the auction.** Parts catalogues and flat-rate labor guides (Mitchell 1946; Audatex 1966; CCC with MOTOR) create a standardized way to turn a damage description into dollars, for the collision repair and insurance industries. 2. **1948–2000s — auctions standardize *disclosure*, not cost.** NAAA (founded 1948, per the "Gold Standard — Since 1948" mark on its guidelines) develops the 0–5 Vehicle Condition Grading Scale and an arbitration policy that references repair cost only as a threshold. 3. **2008 — AutoGrade electronifies the grade and e [source]
21. The salvage-side figure is an insurer's or body shop's retail-basis estimate, which is why it is structurally higher than a rebuilder's actual cost. A buyer's guide states the listed estimate is "written at collision-shop labor rates with new OEM parts, which is precisely why the car was totaled," while "a hands-on rebuilder with used panels routinely repairs for a third of the printed figure." (https://salvagealert.com/guides/acv-repair-estimates-and-lot-numbers-explained) *Lower-confidence: trade/commercial source, no primary backing found for the "one third" ratio. The direction of the [source]
**Claim 3 — The industry-standard auction condition score deliberately excludes repair cost, and this is a documented design decision, not an oversight.** Manheim's AutoGrade service documentation, distributed by NAAA, answers directly: "Q: Does AutoGrade® Condition Score consider the cost of repairs? A: No — Repair labor rates can vary from shop to shop skewing the score. AutoGrade remains consistent by using the damage descriptions." It further states "AutoGrade is not determined by the price of the repairs." The score instead reduces the grade by damage *type and severity*, sorted into Majo [source]
**Claim 8 — Disconfirming evidence: automated damage detection does not yet close the gap to a reliable cost number.** Academic work on vision-based damage detection explicitly stops short of costing. CarDD (Wang, Li and Wu, *IEEE Transactions on Intelligent Transportation Systems* 24(7), July 2023; arXiv preprint submitted 2022-11-02) is presented as "the first public large-scale dataset designed for vision-based car damage detection and segmentation" — 4,000 images, ~9,000 annotated instances, six damage categories — and addresses detection only, with no repair-cost component. The 2025 WIREs [source]
16. For structural defects the auction substitutes a dimensional test for a cost test. NAAA defines the Used Vehicle Measurement Standard (UVMS) as "The commercially acceptable measurement deviation from the vehicle's original structural specification in order for any deviation not to be considered structural damage," and fixes it numerically: "The vehicle structure must measure to a total tolerance of no more than +/- 8 millimeters (mm) of published specification of length, width and height," with side-to-side symmetry within "no more than 6 mm." (https://naaa.com/wp-content/uploads/2026/05/N [source]
1. NAAA / Manheim — AutoGrade® Service: Industry Standard Vehicle Condition Score (retrieved 2026-09-19) — https://www.naaa.com/References/reference_materials/AutoGrade_Information_Document.pdf 2. NAAA Auction Condition Report Position Statement, updated June 1, 2026 — https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html 3. NAAA Arbitration Guidelines, In-Lane and Online, effective June 1, 2026 (14 pp., incl. Appendix I and Structural Component Guidelines) — https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_Jun [source]
**Saturation verdict:** governance layer **saturated** (all four converged on the same primary doc, same sections, same readings — stop searching it). Costing methodology **unsaturated and likely unsaturable** — four independent searches for the damage-code→labor-hour mapping returned zero, and that convergence of negatives is the strongest signal in the run. Accuracy benchmarking for auction vehicles: **nonexistent in the public record**; every figure in that layer is borrowed from collision insurance. Salvage side: thin, two reports, one commercial source for magnitude. [source]
7. **Inference, not a sourced claim:** the seller-submitted repair cost on an insurance-sourced salvage lot is generally the carrier's total-loss decision estimate, written at insurer labour rates with a claims parts mix. A rebuilder's cost basis (recycled parts, own labour) differs in both directions. Copart's own stated reasons for variance — recycled parts, labour rates, paint times — are consistent with this, but no source found states the provenance explicitly. **[Low — reasoned inference, flagged as a gap]** [source]
**Claim 4 — The underlying dollar data comes from the collision-estimating industry, which predates electronic auction condition reporting by roughly half a century.** Mitchell states its origins are "in Glenn Mitchell's garage in 1946," pioneering "the industry's first parts catalogues in 1946," growing into "North America's largest provider of collision repair information." Audatex was founded in Europe in 1966. Together with CCC Information Services (which uses the MOTOR database), these three are the primary parts-and-labor database providers used in the United States, and they supply the [source]
22. Collision estimating databases — the upstream source of the labor times any estimate leans on — carry an assumption that is false for every auction vehicle being estimated. The Database Enhancement Gateway states the databases are "developed and marketed as guides and are based on NEW AND UNDAMAGED PARTS BEING REPLACED ON NEW AND UNDAMAGED VEHICLES." (https://degweb.org/know-your-estimating-system/) [source]
**B9. NAAA is not an enforcement body, so thresholds are defaults, not law.** The guidelines' cover page states they "are not enforced or monitored by NAAA, as the trade association is not a governing or enforcement body," and that "member auctions are free to adopt the NAAA guidelines, in whole or in part, into their contractual relationships at their sole discretion." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) Local addenda do vary the number — DAA of Alaska's NAAA policy addendum applies a $500 single-defect threshold for green-light ve [source]
**G5. For any ADAS-equipped vehicle, add a calibration reserve that the condition report will not have flagged.** F2 and F4 establish that calibration work is both common and systematically missed at estimate time, and V.4.l bars arbitration for ADAS components integrated into visibly damaged parts (B7) — so this cost is both likely and unrecoverable. [source]
Research date: 2026-09-19 Parent domain: Used Car Auctions and Buying Report type: mechanism (internal parts, invariants, limits) [source]
**A2. AutoGrade scores damage by type and severity, not by price, and the grade will not move more for expensive repairs than for cheap ones.** The document states "AutoGrade is not determined by the price of the repairs," and groups damage into Major, Moderate, and Minor grade reductions — Major includes hail, structural damage, mechanical/engine damage, panel replacement, no-keys/non-drivable, and fire/flood/biohazard; Moderate includes glass, wheels, body repair, tire replacement at ≤3/32" tread, interior damage, and damaged/missing parts; Minor includes cosmetic body damage, partial bumper [source]
**E5. Typical magnitudes place the estimate in the same range as the arbitration threshold.** Reconditioning cost per unit is commonly described as ranging roughly $500–$1,500, with ~$1,000 a common ballpark. (https://www.acvauctions.com/blog/auto-reconditioning-for-dealerships) Since the NAAA single-defect threshold is $800 (claim B2), a typical whole-vehicle recon bill sits near the level at which a *single* defect would become arbitrable — meaning most real recon spend is made up of individually non-arbitrable items. [source]
**G1. Treat the $800 / 2% threshold as a modeling parameter, not a footnote.** Because each defect must clear it independently (B3) and wearables are excluded (B7), the expected recoverable share of a recon bill is much smaller than the bill. An estimation model that omits the threshold structure will systematically overstate downside protection. [source]

How it works

This report covers how a repair (reconditioning) cost figure is produced, used, contested, and evaluated for a vehicle transacted at a wholesale auto auction. In scope: the inputs a cost estimate is built from (condition reports, damage line items, grades, diagnostic scans, battery state-of-health), who is authoritative for the number when money turns on it, the dollar thresholds that make an estimate consequential, and how well automated estimation performs. [source]
**Secondary / aggregator (used only where flagged)** - Policygenius, *Total loss threshold by state*, updated 2026-07-20 — https://www.policygenius.com/auto-insurance/total-loss-threshold-by-state/ - WalletHub, *Total Loss Thresholds by State: 2026 Guide* — https://wallethub.com/edu/ci/total-loss-threshold-by-state/104642 - CarInsurance.com, *Total Loss Thresholds* — https://www.carinsurance.com/Articles/total-loss-thresholds.aspx - NIADA Dashboard, *NAAA Sharpens Arbitration Rules on Announcements, High-Value Vehicles, and ADAS* — https://niada.com/dashboard/naaa-sharpens-arbitration-rules-on [source]
- **R1 — patent number settled.** `mechanism.md` flagged NAAA's cited US 8,320,362 as a telecom patent with no resolution. `history.md` independently quoted a *different* NAAA document giving 8,230,362, matching Manheim's 2012 press release. Two of three agree: the transposition is in `Autograde_Std_Vehicle_Condition_Grade_5_15_2018.pdf` alone. - **R2 — the central paradox dissolves.** "The CR disclaims estimating repair cost, yet arbitration turns on a cost figure" was unresolved in `history.md`. `edge-cases.md` supplies the missing artifact: Manheim publishes a separate *Arbitration Validati [source]
It does not cover the parent domain (how auctions work, buying strategy), sibling concepts such as vehicle condition grading as a valuation input, wholesale price forecasting, or title branding. Where a claim touches those, it is included only because the repair-cost estimate is defined in terms of it. [source]
**Claim 3b — The grade itself is not arbitrable; only the underlying undisclosed damage is.** The 2026 Arbitration Guidelines state under "Vehicle Grades & Scoring Systems" that "Auction and seller make no representations or guarantees of scoring systems. Buyers may only arbitrate a vehicle based upon undisclosed damage or defects that were present at the time of the sale." This closes the loop with Claim 3: because the grade carries no dollar figure, it carries no dollar liability. Source: https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf (Section [source]
14. Because grade excludes cost, the grade scale describes reconditioning only qualitatively. The NAAA-recommended 0–5 scale reaches cost language just once, at Grade 1: "It may be cost prohibitive to extensively recondition this vehicle by automotive industry standards." Every other grade is defined by damage description. (https://help.adesa.com/wp-content/uploads/2020/09/vehicle-grading-scale.pdf, p. 2) [source]
- NAAA, *Arbitration Guidelines, In-Lane and Online*, effective 2026-06-01 — https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf - NAAA, *AutoGrade™ — Industry Standard Vehicle Condition Grade*, dated 2018-05-15 — https://naaa.com/wp-content/uploads/2022/11/Autograde_Std_Vehicle_Condition_Grade_5_15_2018.pdf - ADESA, *The NAAA-Recommended Vehicle Condition Grading Scale*, © 2020 — https://help.adesa.com/wp-content/uploads/2020/09/vehicle-grading-scale.pdf - Manheim, *Digital Buyer Protection* marketplace policy — https://site.manheim.com/en/marketpla [source]
**A3. Consequence: the grade is an index of damage, and the buyer must supply the price vector.** AutoGrade's own framing is that the score becomes a price only "when AutoGrade information is combined with mileage, year, make, model, and then used in conjunction with industry valuation tools." (https://www.naaa.com/References/reference_materials/AutoGrade_Information_Document.pdf) The cost estimation step is therefore an explicit, buyer-owned inference layered on top of a deliberately cost-free grading standard. [source]
**Primary / official** - NAAA, *Arbitration Guidelines Update, Effective June 1, 2026* — https://naaa.com/wp-content/uploads/2026/05/Summary_of_Changes_to_Arbitration_Guidelines_and_Position_Statements_June_1_2026.pdf - NAAA, *Arbitration Guidelines, June 1, 2026 (full)* — https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf - Manheim, *NAAA Auction Condition Report Position Statement* — https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html - Manheim, *Arbitration Validation and Estimate Pro [source]
4. **No public specification of how damage lines map to labor hours.** Every source located asserts that the condition report yields labor hours and materials cost, but none publishes the lookup — which damage code at which severity yields how many hours. This is the central mechanism of the concept and it is not documented in the open record. [source]
Met. Sources span nine independent hosts: naaa.com (trade standards body, primary), site.manheim.com and press.manheim.com (marketplace operator, primary policy), help.adesa.com (competing operator, primary), faculty.haas.berkeley.edu (peer-reviewed economics, AER 2015; working paper dated 2014-11-12), degweb.org and scrs.com (collision repair trade bodies), cccis.com (estimating-data vendor, dated 2026-03-31), and support.copart.co.uk / carproxy.com (salvage operator terms). Primary policy documents and one academic paper carry most of the mechanism claims. [source]

Examples and snippets

``` — `~/.global-ai-hub/research-runs/used-car-auctions-20260919/auction-vehicle-repair-cost-estimation/sy` … [source]

Measurements and reference values

20. **AAA's earlier measurement (Oct 2018)** found ADAS behind windshields, bumpers and door mirrors can add up to $3,000 in extra repair cost, with a minor front or rear collision bill running as high as $5,300 — roughly 2.5× the same repair on a vehicle without those systems. Useful as the upper bound on "minor damage" mis-estimation. **[High]** https://newsroom.aaa.com/2018/10/new-vehicle-technologies-double-repair-bills-minor-collisions/ [source]
**B2. The arbitration threshold is $800 for vehicles under $50,000 and 2% of purchase price at or above $50,000.** Section V.2.b: "For vehicles with a purchase price under $50,000, the repair cost for any single undisclosed or unannounced mechanical defect to be eligible for arbitration is $800 or more"; "For vehicles sold at a purchase price of $50,000 or greater, the repair cost for any single undisclosed or unannounced mechanical defect to be eligible for arbitration is two percent (2%) of the purchase price or more." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_ [source]
**B5. Disclosure thresholds on the seller side use the same $800 / 2% figures.** Appendix I footnote: "Defects with a repair or replacement cost of $800 or more must be disclosed (2% threshold applies to vehicles sold for $50,000 or more)." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) The seller is therefore also running a repair cost estimate, pre-sale, against the identical cutoff — and the two estimates are made by different parties with opposed incentives. [source]
**D2 — ADAS share of the repair bill: ceiling quoted as central tendency.** AAA's finding is that ADAS *can add up to* 37.6% of total repair cost on the vehicles tested (https://newsroom.aaa.com/2023/12/fixing-advanced-vehicle-systems-makes-up-over-one-third-of-repair-costs-following-a-crash/), whereas trade coverage renders it as ADAS "making up over a third of collision repair costs" generally (https://www.autobodynews.com/news/fixing-adas-systems-makes-up-over-a-third-of-collision-repair-costs). A worst-case ceiling on selected ADAS-equipped vehicles is not an average across the auction fle [source]
**D5 — Whether AI photo estimation transfers from claims to auction lots is untested.** The ~84% capture figure (claim 25) is described in insurer/DRP claim workflows on repairable vehicles with a known loss event. Auction lots — especially salvage lots with unknown history, prior repairs, and no claim file — are a different distribution. No independent evaluation of photo-based estimation accuracy on auction inventory was found. [source]
**Claim 2 — Despite Claim 1, a repair cost estimate is the legal trigger for auction arbitration, and the auction itself is the estimator of record.** The NAAA Arbitration Guidelines effective June 1, 2026 set dollar thresholds a defect must exceed before a buyer can arbitrate: "For vehicles with a purchase price under $50,000, the repair cost for any single undisclosed or unannounced mechanical defect to be eligible for arbitration is $800 or more," and "For vehicles sold at a purchase price of $50,000 or greater, the repair cost for any single undisclosed or unannounced mechanical defect to [source]
17. **About half of calibration lines are discovered after the vehicle is opened up.** CCC's Q4 2025 Crash Course reports that 51.5% of Q3 2025 calibrations appeared on a supplement, with 48.5% on the initial estimate (E01) — while 92.1% of scans appeared on the E01. A photo-based or CR-based pre-purchase estimate therefore misses roughly half of calibration cost by construction, even when scanning is captured up front. **[High]** https://www.cccis.com/reports/crash-course-2025/q4 [source]
18. **Calibration unit cost, dated:** Enlyte/Mitchell 2026 Envision Trends reports the average cost of a calibration reached US$688 when present, calibration lines grew 31.4% year over year in 2025, and calibrations per repair rose nearly 10% from 2024 when present at all. **[Med — trade reporting of a vendor dataset]** https://www.repairerdrivennews.com/2026/06/04/calibrations-grew-31-4-in-2025-mitchell-data-shows/ [source]
19. **AAA's measured ceiling (Dec 2023): ADAS can add up to 37.6% to total repair cost after a crash**, with minor damage to front radar or distance sensors adding up to $1,540. **[High]** https://newsroom.aaa.com/2023/12/fixing-advanced-vehicle-systems-makes-up-over-one-third-of-repair-costs-following-a-crash/ [source]
**D6. EV battery state of health is the one condition input that auctions now measure directly and publish per-VIN.** Cox Automotive announced on March 27, 2024 a VIN-specific battery health solution deployed across Manheim locations, using LotVision to "quickly assess the battery in each unique vehicle and automatically transmit in-depth battery health information to condition reports (CRs) and vehicle detail pages (VDPs) on Manheim.com," reporting a 0–100% score representing remaining capacity captured directly from the vehicle. (https://press.manheim.com/2024-03-27-Cox-Automotive-Deploying- [source]
10. **Per-defect minimums, effective June 1, 2026 (NAAA):** for vehicles sold under $50,000 the $800 per-defect threshold is unchanged; for vehicles sold at $50,000 or more each individual arbitrable defect must carry a repair cost of at least 2% of purchase price. Worked examples in the policy: $50,000 vehicle → $1,000 minimum; $75,000 → $1,500; $100,000 → $2,000. **[High]** https://naaa.com/wp-content/uploads/2026/05/Summary_of_Changes_to_Arbitration_Guidelines_and_Position_Statements_June_1_2026.pdf [source]
**Dated technical reports and data** - CCC Intelligent Solutions, *Crash Course 2026* release, 2026-03-31 — https://www.cccis.com/news-and-insights/posts/ccc-crash-course-2026-report-finds-higher-severity-and-record-total-loss-frequency - CCC Intelligent Solutions, *Q4 2025 Crash Course* — https://www.cccis.com/reports/crash-course-2025/q4 - AAA Newsroom, *Fixing Advanced Vehicle Systems Makes Up Over One Third of Repair Costs Following a Crash*, Dec 2023 — https://newsroom.aaa.com/2023/12/fixing-advanced-vehicle-systems-makes-up-over-one-third-of-repair-costs-following-a-crash/ - AAA Newsroom [source]
**Claim 7 — Disclosing condition information at wholesale auction measurably raises prices, which is the economic justification for the whole inspection apparatus.** Tadelis and Zettelmeyer, "Information Disclosure as a Matching Mechanism: Theory and Evidence from a Field Experiment," *American Economic Review* 105(2), 2015, 886–905, ran a large-scale field experiment randomly disclosing quality information in wholesale automobile auctions. The abstract states: "information disclosure helps match heterogeneous buyers to cars of varying quality, causing both good and bad news to increase compet [source]
8. The same threshold runs in the other direction as a seller duty. NAAA's Appendix I footnote reads: "Defects with a repair or replacement cost of $800 or more must be disclosed (2% threshold applies to vehicles sold for $50,000 or more)." The estimate is therefore the trigger for announcement as well as for arbitration. (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, Appendix I, p. 10) [source]
9. Marketplace-level protection products add a per-defect floor on top of the aggregate floor. Manheim's Digital Buyer Protection states that for vehicles under $50,000 "A minimum total claim value of $800 is required" and "Each cosmetic or structural defect must meet a minimum estimated wholesale repair cost of at least $400"; for vehicles at or above $50,000 the same page sets 2% of purchase price aggregate and 1% of purchase price per defect. (https://site.manheim.com/en/marketplace-policies/us-policies/digital-buyer-protection.html) [source]
24. **Age concentration compounds this:** vehicles seven to twelve years old made up nearly 41% of total-loss valuations in 2025, up from 33.4% in 2020, and average US light-vehicle age reached 12.8 years in 2025. Labour rates do not fall with vehicle age, so minor cosmetic damage totals an old car. Edge case: auction lots described as heavily damaged may be cheaply repairable; lots described as lightly damaged may still have been totalled for ADAS or structural reasons. **[Med]** https://www.partstrader.com/predicting-claims-volume-in-2026-key-factors-to-watch/ [source]
28. **More accurate condition disclosure does not simply lower prices on bad cars.** Tadelis and Zettelmeyer (AER 2015; working paper dated 2014-11-12) ran a randomised field experiment disclosing inspection-based condition information in wholesale automobile auctions, in a market selling thousands of vehicles weekly at an average value of $8,500. Revenues rose when the disclosed information was a surprise in *either* direction — good or bad — and were unchanged when it matched prior expectations, because buyers are horizontally differentiated and sort into the auctions matching their preferre [source]
**D4 — Whether teardown/blueprinting materially reduces cost surprises is unverified.** A trade source claims tear-down estimates cut supplement rates from an industry ~30% average to below 15% (https://www.fleetmanagementweekly.com/supplements-minimize-nasty-collision-repair-surprises/), and a shop blog attributes "63% of collision repairs require at least one supplement" to CCC 2024 data (https://nationalcollisionauthority.com/supplement-process-in-collision-repair/). Neither figure was traceable to CCC's own published reports in this run; CCC's Q4 2025 report states only that the supplement [source]
**F6. Academic automated repair-cost estimation reports strong in-sample metrics on narrow, synthetic-cost datasets.** A July 20, 2026 *Scientific Reports* study (DOI 10.1038/s41598-026-61488-5) pairs YOLOv8 damage detection with XGBoost cost regression, reporting detection precision 0.87 (validation) / 90.7% (test), [email protected] of 0.824, [email protected]:0.95 of 0.606, and for cost regression R² = 0.9728 with MAE of 128.50 EGP, or 7.66% of mean cost. (https://pmc.ncbi.nlm.nih.gov/articles/PMC13385360/) [source]
16. **Calibration is now on roughly a third of repairable estimates and rising fast.** CCC reports calibrations appeared on 28.3% of all repairable appraisals in 2025, up from 21.8% — a ~30% increase in estimates containing at least one calibration. **[High]** https://www.cccis.com/news-and-insights/posts/ccc-crash-course-2026-report-finds-higher-severity-and-record-total-loss-frequency [source]
Primary — peer-reviewed - Tadelis, S. and Zettelmeyer, F., "Information Disclosure as a Matching Mechanism: Theory and Evidence from a Field Experiment," *American Economic Review* 105(2), 2015, 886–905 — https://www.aeaweb.org/articles?id=10.1257%2Faer.20110753 - Working paper full text — https://faculty.haas.berkeley.edu/stadelis/autoauction_073114.pdf - Wang, X., Li, W. and Wu, Z., "CarDD: A New Dataset for Vision-based Car Damage Detection," *IEEE Transactions on Intelligent Transportation Systems* 24(7), July 2023 — https://arxiv.org/abs/2211.00945 - Hasan et al., "Vehicle Damage Detectio [source]
7. The estimate exists primarily to be tested against a disclosure and arbitration threshold. NAAA sets it at "$800 or more" for vehicles with a purchase price under $50,000, and "two percent (2%) of the purchase price or more" for vehicles sold at $50,000 or greater. (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §V.2.b) [source]

Problems, failure modes and limitations

This report covers only the estimation of repair and reconditioning cost for a vehicle being bought or sold through an auction channel — wholesale dealer auctions governed by NAAA arbitration rules (Manheim and peers) and salvage/insurance auctions (Copart, IAA). It covers where those estimates break, what they structurally cannot see, and where published sources disagree. It does not cover vehicle valuation, bidding strategy, title branding, transport, or auction selection; those are separate frontier items. [source]
**Claim 1 — The auction condition report is explicitly not a repair cost estimate.** The NAAA Auction Condition Report Position Statement, updated June 1, 2026, states: "ACRs are not intended to diagnose vehicle condition, certify safety or roadworthiness, or estimate repair costs." It adds that "visual inspections are limited in nature and cannot reliably identify mechanical, ADAS, electrical, structural, or battery-related conditions." The condition report is therefore a damage *disclosure* document, not a costing document. [source]
**Claim 2a — The cost basis is wholesale, not retail, and is deliberately narrowed.** The same guidelines limit successful-arbitration reimbursement to "commercially reasonable costs based on local auction (wholesale) repair or transportation rates," and explicitly exclude reconditioning: "Lost profit and commissions along with marketing and reconditioning costs are not reimbursable under any circumstances." Auction repair cost estimation is thus a distinct cost standard from collision-shop or insurance estimating, which price at retail labor rates. Source: https://naaa.com/wp-content/uploads/ [source]
**B6. Reconditioning cost is explicitly non-recoverable, so an estimating error is borne entirely by the buyer.** Buyer Responsibilities IV.2: "Following a successful arbitration claim, the buyer will be responsible for, and will not be reimbursed by the seller or auction, any repairs, reconditioning, or enhancements made to the vehicle prior to its return to the auction." Seller Responsibilities III.10 further limits reimbursement "to commercially reasonable costs based on local auction (wholesale) repair or transportation rates," and states "lost profit and commissions along with marketing a [source]
**F5. Estimation error also drives the total-loss boundary, which is itself at a record.** CCC reported a 22.8% total loss rate through October 2025, up from 22.1% year over year, with average repair cost of $4,768 through Q3 2025 (up 1.4%). (https://www.autobodynews.com/news/calibrations-surge-past-35-of-repairs-as-total-losses-head-toward-second-straight-record) CCC's 2026 Crash Course reports a 23.1% total loss rate. (https://ir.cccis.com/news-releases/news-release-details/ccc-crash-course-2026-report-finds-higher-severity-and-record) Near that boundary, a modest estimating error flips the [source]
**D8. The arbitration rules treat EV battery problems as a disclosure item with a specific age cutoff.** Appendix I lists "Electric or Hybrid Vehicle Battery Problem (up to eight (8) model years)" as a Major Component requiring green-light disclosure, and separately requires disclosure of "non-original engine or electric vehicle batteries for current model year and up to 4 model years old." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) Beyond eight model years the estimation risk reverts fully to the buyer. [source]
5. **Copart's "Repair Cost" is a seller-submitted number carrying an explicit variance warning.** Copart's glossary defines it as the estimated cost to repair as submitted by the seller, states it "may vary greatly from the actual cost to repair this vehicle," and attributes the variance to availability and use of recycled parts, varying labour rates, and paint times. Copart makes no representation as to its validity or accuracy. **[Med — direct page fetch returned empty; text obtained via search-engine rendering of the same Copart support URL]** https://support.copart.co.uk/faq/what-do-copart [source]
8. **Manheim arbitration estimates are priced at fixed labour rates:** $95.00 per labour hour for mechanical ($135.00 in California) and $75.00 per labour hour for paint and body ($125.00 in California), and "will reflect local auction/wholesale cost to repair." Failure mode: a retail body-shop quote can clear a threshold that the auction's own wholesale-rate estimate does not. **[High]** https://site.manheim.com/en/marketplace-policies/us-policies/arbitration-estimate-process.html [source]
**Known limitation of this report.** Claim F1 stands: no independent, published benchmark of auction repair-cost-estimate accuracy against realized repair invoices was located. Every accuracy figure in Section F is borrowed from an adjacent domain (collision insurance) or from academic work on a non-auction dataset, and is labeled as such. A reader should not treat Section F as measuring auction estimation directly. [source]
15. **ADAS components are now formally in NAAA Appendix I — with a carve-out that shifts cost to the buyer.** If an ADAS component is housed within a visibly damaged part (bumper, mirror, windshield) or was disclosed as damaged in the listing or condition report, it is not subject to arbitration. Only if the ADAS component fails and the associated part was neither damaged nor disclosed does standard eligibility apply. Failure mode: a disclosed cosmetic bumper scuff silently transfers sensor replacement and calibration cost to the buyer. **[High]** (claim 10 URL) [source]
26. **Named feedback-loop failure mode:** AI estimates trained on historical estimate data inherit the bias of that data — lowballed estimates in the training database lead the model to lowball future claims. Relevant because auction-side automated estimators would be trained on insurer/DRP estimates written under cost pressure. **[Low — trade commentary]** https://www.repairerdrivennews.com/2017/10/16/ccc-to-use-ai-photo-estimating-to-classify-total-losses-from-a-single-photo/ [source]
15. A buyer cannot arbitrate the grade; only the underlying undisclosed damage. NAAA states under "Vehicle Grades & Scoring Systems" that "Buyers may only arbitrate a vehicle based upon undisclosed damage or defects that were present at the time of the sale," and the ADESA-published NAAA scale carries the line "VEHICLE CONDITION GRADE CANNOT BE USED FOR ARBITRATION." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §V.4.h; https://help.adesa.com/wp-content/uploads/2020/09/vehicle-grading-scale.pdf, p. 2) [source]
**D7. Cox's stated rationale is that battery degradation is a fat-tailed risk that per-model averages cannot price.** The announcement contrasts its approach with "the grading systems of other EV battery diagnostic tools, which provide a general battery estimate based on make and model," and notes that "while most used EV batteries are relatively healthy, the outliers tend to be significantly degraded." (https://press.manheim.com/2024-03-27-Cox-Automotive-Deploying-Industry-First,-VIN-Specific-EV-Battery-Health-Solution-Across-Manheim-Locations) [source]
2. **The detection boundary is explicitly visual.** The same statement: "Visual inspections are limited in nature and cannot reliably identify mechanical, ADAS, electrical, structural, or battery-related conditions, including those affecting Electric and Hybrid vehicles." This is the hard ceiling on pre-purchase estimation accuracy in the wholesale lane. **[High]** (same URL as claim 1) [source]
11. **Defects cannot be aggregated to clear the threshold.** The policy states the 2% calculation is applied per individual defect and "each must independently meet or exceed the minimum"; multiple defects may be filed together but each must clear its own threshold. Failure mode: a vehicle with five separate $600 defects — $3,000 of real cost — is not arbitrable at any price tier. **[High]** (same URL) [source]
13. **Estimate-relevant disclosure loopholes were closed in the same update.** Generic "No Arb for [defect]" announcements are now prohibited and may themselves be grounds for arbitration with the sale voided; announcing only a warning light or a diagnostic trouble code does not relieve the seller of arbitration responsibility. Before this, a vague announcement could strand a buyer's post-purchase repair cost. **[High]** (same URL) [source]
25. NAAA expressly forecloses arbitrating the ADAS systems integrated into visibly damaged parts, which caps the auction estimate's exposure to exactly the cost class in claim 24: "The buyer cannot file an arbitration claim for ADAS systems and components that are integrated into parts (e.g., bumpers, mirrors, windshields, etc.) which are visibly damaged or were disclosed/announced at time of sale as damaged." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §V.4.l) [source]
**F3. This benchmark sets a hard ceiling on auction-CR-based estimates.** If a licensed estimator with the vehicle physically present misses half of all required calibrations, an estimate built from 30–40 photographs and an OBD2 scan (D1) cannot plausibly do better on the same category of work. This is the strongest available disconfirming evidence against treating a CR-derived number as a firm cost. [source]
**B3. Defects cannot be aggregated to clear the threshold — each must independently qualify.** Section V.2.c.i permits multiple defects in one claim "provided that each individual defect independently meets the arbitration threshold defined in section 2(b) above." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) An independent practitioner summary reaches the same reading: "Each defect must independently meet the threshold; smaller issues cannot be combined." (https://getcarbly.com/blog/what-2026-naaa-arbitration-changes-mean-before-you-bid-remo [source]
12. **One mechanical filing per transaction.** A buyer gets one opportunity to file a mechanical arbitration claim per vehicle transaction; multiple mechanical defects must be bundled into that single filing. Structural-damage and vehicle-history claims are separate and are not restricted by a prior mechanical filing. Failure mode: filing on the first discovered defect before teardown forfeits later mechanical discoveries. **[High]** (same URL) [source]
**Claim 2b — Each defect is costed separately; defects cannot be aggregated to clear the threshold.** The guidelines allow a buyer one arbitration opportunity per vehicle for mechanical defects and permit multiple defects in that claim, "provided that each individual defect independently meets the arbitration threshold defined in section 2 (b) above." Source: https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf (Section V, item 2.c.i) [source]
**Claim 8a — Disconfirming evidence from practice: photo-based estimates systematically miss work that surfaces only on teardown.** CCC's data, reported by Repairer Driven News, found that only 45.5% of ADAS calibrations appeared on the initial estimate; the remaining 54.5% were added later as supplements once the shop knew what the vehicle actually needed. This is the practical mechanism behind Claim 1's warning that visual inspection cannot reliably identify ADAS, electrical or structural conditions. [source]
24. An undamaged-state visual inspection cannot see the damage classes that now dominate cost growth. CCC reports that "28.3% of repairable estimates now include calibrations" for camera- and sensor-based ADAS, up from 21.8% a year earlier, and that "Total loss frequency reached 23.1% of claims, a new industry high." (https://www.cccis.com/news-and-insights/posts/ccc-crash-course-2026-report-finds-higher-severity-and-record-total-loss-frequency, dated 2026-03-31) [source]
**A5. The same statement bounds what a visual inspection can detect, which bounds what any estimate built on it can cover.** It states that "visual inspections are limited in nature and cannot reliably identify mechanical, ADAS, electrical, structural, or battery-related conditions, including those affecting Electric and Hybrid vehicles," and that identifying such conditions "typically requires trained technicians and specialized equipment." (https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html) [source]
**F4. The share of vehicles where this failure mode applies is large and growing.** Calibrations appeared on 35.6% of DRP estimates in Q3 2025, up from 26.9% a year earlier; diagnostic scans appeared on 87.7%, up from 84.7%. (https://www.autobodynews.com/news/calibrations-surge-past-35-of-repairs-as-total-losses-head-toward-second-straight-record) [source]
**G2. Estimate in two separated components: an arbitrable tail and a non-arbitrable base.** The base (wearables, cosmetics, detail, items under threshold) is a certain cost to be bid down for. The tail (undisclosed structural, engine/transmission, EV battery within 8 model years) is a contingent cost with a partial recovery path. Only the tail benefits from arbitration; conflating the two is the error B6 punishes. [source]

Comparisons and alternatives

23. **Disconfirming: record total-loss frequency is substantially an ACV-denominator effect, not a repair-cost-numerator effect.** CCC reports total loss frequency hit a record 23.1% of claims in 2025 while noting 12 million fewer vehicles aged six years or newer in operation as of Q3 2025 versus 2020. Average total cost of repair grew only from $4,700 to $4,768 through Q3 2025 (+1.4% YoY) — near-flat. A buyer modelling auction supply purely from repair-cost inflation misreads the mechanism. **[High]** (claim 16 and claim 17 URLs) [source]
**Claim 5 — In salvage auctions the repair cost estimate is a published listing field and the economic pivot of the transaction, unlike in dealer auctions.** IAA lot listings expose "actual cash value" and "estimated repair cost" alongside primary and secondary damage, loss type, run-and-drive and keys status. IAA's Loss Advisor product is built around the same arithmetic: the reported damage, the ACV and the salvage value together determine whether a vehicle is a total loss. This is the inverse of the dealer-auction convention in Claim 1 — the salvage side publishes a cost number because the [source]
**B7. Whole categories of cost are structurally excluded from arbitration, so they must be priced into the bid rather than recovered later.** Section V.4 excludes wearable items (tires, wipers, belts, brake pads, rotors, bulbs, filters, timing belts, shocks, struts, air ride, suspension components, fluids), vehicles over 20 model years, inherent conditions typical to a model, manual clutch assemblies, non-actively-dripping oil leaks, vehicle accessories on vehicles over 4 model years old, and ADAS components integrated into visibly damaged or disclosed parts. (https://naaa.com/wp-content/uploa [source]
**D1. Auction inspection has moved toward instrumented capture rather than toward supplying a cost number.** ACV's condition reports "include 30-40 photographs, OBD2 scans, paint meter readings, and tire tread depths." (https://www.acvauctions.com/blog/no-more-surprises-how-condition-reports-make-a-difference) [source]
**Gaps the gate does not close:** no peer-reviewed or primary measurement of estimated-versus-actual repair cost *specifically for auction-purchased vehicles* was located. Claims 25, 26 and D4 rest on vendor or trade material only. The Copart disclaimer text (claims 5–6) could not be retrieved by direct page fetch — both support.copart.co.uk and copart.ie returned empty bodies — and is recorded at medium confidence from search-engine rendering of the same official URL. [source]
**F1. There is no published, independent accuracy benchmark for auction repair cost estimates.** No source located for this report reports error distributions for auction-context recon estimates against realized repair invoices. Auction operators publish inspection capability claims (D1–D6) but not estimate-versus-actual calibration. This is a genuine measurement gap, not merely an unsearched one — see Unresolved Disagreements. [source]
**F7. That study's R² is not transferable to the auction setting, for four concrete reasons.** Its cost labels were "created from scratch with the help of repair experts" rather than drawn from realized invoices; it covers only five cosmetic damage classes (dent, crack, scratch, broken lamp, glass shatter); its prices are denominated in Egyptian pounds; and it reports no external validation against independent insurance or repair-shop data. (https://pmc.ncbi.nlm.nih.gov/articles/PMC13385360/) A high R² against expert-generated labels measures agreement with the labelers, not with the eventual [source]
**U1. "Consistency" versus "usefulness" in the grading standard.** AutoGrade's exclusion of repair cost is justified as removing labor-rate variance and thereby increasing cross-auction consistency (A1). The cost of that choice is that the score carries no price information and the buyer must reconstruct it (A3). No source located quantifies whether the consistency gain outweighs the estimation burden it creates. Both framings are defensible; the trade-off appears not to have been measured publicly. [source]
**Disconfirming sources actively sought and found.** Three distinct disconfirming lines are incorporated rather than set aside: (a) the AutoGrade documentation explicitly refutes the intuitive assumption that condition grades encode repair cost (A1–A2); (b) CCC's supplement data refutes the assumption that expert in-person estimation is reliable, which caps what remote estimation can achieve (F2–F3); (c) the *Scientific Reports* limitations refute the transferability of headline automated-estimation accuracy to this setting (F7). [source]
21. **BEVs carry more calibration exposure than ICE**: 1.70 calibrations per estimate versus 1.54 for ICE and 1.63 for hybrids, while US BEV claim cost actually *fell* 5% in 2025, from $6,707 to $6,395. Calibration count and total severity move independently — a partly disconfirming data point for "EVs are uniformly more expensive to estimate." **[Med]** https://www.mitchell.com/insights/article/auto-physical-damage/plugged-in-ev-collision-insights-2025-year-review [source]
**C1. Structural disputes are resolved against a dimensional standard rather than a dollar estimate.** The Structural Component Guidelines define UVMS (Used Vehicle Measurement Standard) as "the commercially acceptable measurement deviation from the vehicle's original structural specification in order for any deviation not to be considered structural damage," and set the tolerance at "no more than +/- 8 millimeters (mm) of published specification of length, width and height at control points," with symmetrical point-to-point measurements differing by "no more than 6 mm." (https://naaa.com/wp-c [source]
**D4. Tire measurement is reported at 1/32" resolution by automated inspection towers.** ACV's VIPER "automatically scans all 4 tires from multiple angles to accurately detect damage and see detailed tread depth measurements to the nearest 1/32"." (https://investors.acvauto.com/news-events/press-releases/detail/102/acv-demonstrates-viper-ai-powered-inspection-towers-at-the-2026-national-automobile-dealers-association-conference) This resolution matters because AutoGrade's tire bands are defined at exactly that granularity (≤3/32" versus 4/32"–7/32"). (https://www.naaa.com/References/reference_ [source]
22. **The total-loss trigger varies enormously by jurisdiction.** Policygenius (updated 2026-07-20) lists Oklahoma lowest at 60% and Colorado and Texas at 100%, with 75% the most common percentage, and lists 22 states plus several others using the Total Loss Formula (ACV vs. repair cost + salvage value) instead of a flat percentage. The same physical damage and the same repair estimate produce a salvage lot in Oklahoma and a repaired car in Colorado. **[Med — aggregator, see disagreement D3]** https://www.policygenius.com/auto-insurance/total-loss-threshold-by-state/ [source]
3. **Severity-based vs cost-based scoring.** AutoGrade argues cost is too variable to score on (Claim 3). Salvage auctions publish a cost number anyway (Claim 5). Both are defensible; the sources do not engage each other, and no comparative study was located. [source]
This report covers how a dollar figure for repairing a vehicle is produced and used inside an auction transaction. It covers the inputs (an inspector's damage line items), the arithmetic (labor hours times a fixed wholesale labor rate, plus parts and materials), the rate and parts conventions that make the figure "wholesale" rather than retail, and the dollar thresholds the figure is compared against. It covers both wholesale dealer auctions (Manheim, ADESA, NAAA member auctions) and salvage auctions (Copart), because both attach a repair-cost number to a lot. [source]
25. **AI photo estimating is reported to capture roughly 84% of final estimate value** in preliminary estimates produced in under 80 seconds, with 95–97% claimed certainty on the repair-versus-total-loss classification. The ~16% shortfall is the pre-teardown gap, consistent with claim 17. **[Low — vendor marketing figures relayed by a trade blog; no independent evaluation found]** https://www.wickedfile.com/blogs/how-ai-is-revolutionizing-auto-body-repair-the-shift-you-cant-ignore/ [source]
4. **Pre-reconditioning CR versions differ from the sale CR.** Manheim shares pre-reconditioning report versions with sellers for repair estimating and remarketing; these "may differ from the final report available at sale." Failure mode: comparing a seller-side recon estimate against a buyer-side CR treats two different documents as one. **[Med]** https://site.manheim.com/en/services/condition-reporting.html [source]
**D1 — Calibration prevalence: 28.3% vs 34.7% vs 35.6%.** CCC's 2026 annual report says 28.3% of *all repairable appraisals* in 2025 included a calibration (https://www.cccis.com/news-and-insights/posts/ccc-crash-course-2026-report-finds-higher-severity-and-record-total-loss-frequency). CCC's own Q4 2025 quarterly says 35.6% of Q3 2025 *repairable DRP appraisals* included one (https://www.cccis.com/reports/crash-course-2025/q4). Enlyte/Mitchell reports 34.7% of 2025 repair estimates (https://www.collisionrepairmag.com/multi-media/podcast/repair-trends/article/15826600/mitchell-international-ad [source]
**F2. Professional, in-person collision estimating — a much better-instrumented analogue — still misses a large share of work at first pass.** CCC data reported for Q3 2025 shows only 48.5% of calibrations appear on initial estimates, with 51.5% added on supplements, versus 92.1% of diagnostic scans captured on initial estimates. (https://www.autobodynews.com/news/calibrations-surge-past-35-of-repairs-as-total-losses-head-toward-second-straight-record) CCC attributes this to "the challenge identifying all of the needed repairs at the time the estimate is generated, especially if no tear down i [source]
**G6. For EVs, use the per-VIN state-of-health score, and widen the estimate sharply when it is absent or the vehicle is over 8 model years old.** D6 makes the measurement available; D7 establishes that the risk is in the tail rather than the mean; D8 removes the disclosure backstop past 8 model years. [source]
**G8. Do not expect a grade-to-dollar lookup table to exist as a published standard.** A1–A3 establish that no such mapping is published by NAAA or AutoGrade, because the standard was designed to exclude cost. Any grade-to-dollar table in use is a private, dealer-specific fit, and should be validated against that dealer's own realized recon data rather than adopted from elsewhere. [source]
**U6. Local variation in thresholds is documented but not characterized.** B9 shows at least one auction applying $500 rather than $800, but no source located surveys how widely local addenda deviate from the NAAA defaults. A buyer operating across multiple auctions cannot assume a single threshold. [source]

Facts and statements

**Run:** used-car-auctions-20260919 · **Concept:** Auction Vehicle Repair Cost Estimation **Parent context:** Used Car Auctions and Buying · **Date:** 2026-09-19 [source]
Run: `used-car-auctions-20260919` · Concept: Auction Vehicle Repair Cost Estimation · Parent: Used Car Auctions and Buying · Compiled 2026-09-19 [source]
**Concept:** Auction Vehicle Repair Cost Estimation **Parent context:** Used Car Auctions and Buying **Report type:** practice (operational use, trade-offs, evaluation, implications) **Date written:** 2026-09-19 [source]
1. **Auction condition reports formally disclaim repair-cost estimation.** The NAAA Auction Condition Report position statement states condition reports "are not intended to diagnose vehicle condition, certify safety or roadworthiness, or estimate repair costs." Any repair-cost figure derived from a CR is the buyer's inference, not a representation by the auction. **[High]** https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html [source]
This report covers the estimation of the cost to repair or recondition a vehicle that is being transacted at auction — wholesale dealer-to-dealer auctions (Manheim, ADESA, ACV) and salvage auctions (IAA, Copart). It covers who produces the estimate, what the estimate is used for (bidding, arbitration, total-loss decisions), what data it rests on, and how the practice has changed over time. [source]
**D3. Engine condition is captured as an audio signal scored by machine learning, not as a repair estimate.** ACV's Audio Motor Profile (AMP) uses "patented and proprietary Audio Motor Profile (AMP™) recordings" with "machine learning models" to capture vehicle condition. (https://www.acvauctions.com/blog/auto-reconditioning-for-dealerships) [source]
**D5. Manheim's enhanced condition report added visualization and damage heat maps, not cost output.** The InSight Condition Report announcement describes manufacturer data integration, a digital walk-around, pan-and-zoom photography, heat map highlighting of damage areas, and historical CR data — with no repair or reconditioning cost estimate among its outputs. (https://press.manheim.com/InSightCR) [source]
Out of scope, and deliberately not researched: how auctions are run, how buyers bid, how vehicles are valued or priced, condition grading as a topic in its own right, total-loss branding law, and collision repair as a trade. Condition grading and collision estimating databases appear here only where they are inputs to or explicit non-inputs of the repair cost figure. [source]
1. The auction repair cost estimate is computed from an inspector's damage line items, not from a shop's quote. The standardized condition report "calculates the expected number of labor hours needed for a body-shop technician to correct the reported damage, as well as the cost of the materials needed. Using a standard hourly labor rate, this translates into the cost of bringing the vehicle to a condition where exterior and interior damage are no longer noticeable." (https://faculty.haas.berkeley.edu/stadelis/autoauction_073114.pdf, p. 5) [source]
19. At salvage auctions the repair cost figure is not computed by the auction at all; it is passed through from the seller. Copart defines it as "the estimated cost to repair the vehicle as submitted to Copart by the Seller of the vehicle," and defines Estimated Retail Value the same way: "the estimated retail value of the lot as submitted to Copart by the Seller." (https://carproxy.com/how-it-works/common-copart-terms; https://support.copart.co.uk/faq/what-do-copart-terms-mean/) [source]
**A4. Auction condition reports carry an explicit disclaimer that they do not estimate repair cost.** The NAAA Auction Condition Report Position Statement, updated June 1, 2026, states that "ACRs are not intended to diagnose vehicle condition, certify safety or roadworthiness, or estimate repair costs." (https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html) [source]
27. **CCC also markets AI classification of total losses from a single photo**, which is a classification claim, not a cost-estimation claim; conflating the two overstates what photo-based tools deliver for repair *cost*. **[Med]** (same URL as claim 26) [source]
**B4. The buyer gets one mechanical arbitration attempt per vehicle.** Section V.2.c: "A buyer is only allowed one opportunity in time per vehicle purchase to file an arbitration claim for mechanical defects." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) This creates a timing problem for estimation: filing on the first discovered defect forfeits later-discovered ones, so the buyer must complete a full mechanical assessment before filing. [source]
**D2. Undercarriage imaging is now a discrete product, addressing a region the standard walk-around misses.** ACV's Virtual Lift is described as "the industry's first mobile vehicle undercarriage imaging tool," producing "a full bumper-to-bumper view and image of the vehicle's undercarriage in less than one minute." (https://www.prnewswire.com/news-releases/acv-auctions-unveils-virtual-lift-300893699.html) [source]
9. **The arbitration cost report is itself not a repair estimate.** Manheim states ACRs are not intended to diagnose condition, certify roadworthiness, or estimate repair costs, and that an ACR "is not a warranty, guarantee, mechanical certification, or representation of future vehicle performance." **[High]** (claim 1 URL) [source]
14. **Unannounced pre-sale defects that will produce a future title brand are arbitrable** under the reorganised Title Policy, with DMV records, insurance declarations, law-enforcement reports and prior auction history named as acceptable documentation. This is a cost channel that sits outside any physical repair estimate. **[High]** (same URL) [source]
1. **Is the auction condition report a repair cost estimate or not?** NAAA says explicitly it is not (Claim 1), yet arbitration turns entirely on a repair cost figure that the auction itself produces (Claim 2). The position statement disclaims the estimate while the arbitration guidelines require one. Unresolved in the sources: whether the arbitration estimate is produced from the condition report at all, or from a separate inspection at claim time. Section V.2.d ("The auction will only inspect the mechanical defect(s) on the initial arbitration claim") implies a separate inspection, but no so [source]
2. **What method does an auction use to compute "local auction/wholesale cost to repair"?** The guidelines name the standard but publish no methodology, no rate table, and no requirement to use any of the three estimating databases. No source located states whether auctions run Mitchell/CCC/ Audatex estimates at wholesale rates, use internal recon-shop pricing, or use a mix. This is the single largest documentation gap for this concept. [source]
6. The auction, not the buyer or the seller, is the authority that sets the number in a dispute. NAAA's guidelines state: "For purposes of the dollar thresholds for arbitration, the repair cost will be determined by the auction based upon the local auction/wholesale cost to repair the defect(s)." The same section adds that "The auction decision on all arbitration claims is final and binding for both the buyer and seller." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §V.2.d and §V.2.f) [source]
10. Reimbursement is capped at the same wholesale basis the estimate was written on. NAAA states buyer reimbursement will be "limited to commercially reasonable costs based on local auction (wholesale) repair or transportation rates," and that "Lost profit and commissions along with marketing and reconditioning costs are not reimbursable under any circumstances." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §III.10) [source]
27. Once the vehicle leaves the auction, the buyer owns every repair dollar regardless of the estimate. NAAA: "Following a successful arbitration claim, the buyer will be responsible for, and will not be reimbursed by the seller or auction, any repairs, reconditioning, or enhancements made to the vehicle prior to its return to the auction," excepting undisclosed conditions "not detectable through vehicle inspection." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §IV.2) [source]
**A1. The industry-standard condition score deliberately excludes repair cost from its calculation.** The AutoGrade service, which supplies the 0–5 condition score used across North American wholesale auctions, answers the question directly: "Does AutoGrade® Condition Score consider the cost of repairs? A: No – Repair labor rates can vary from shop to shop skewing the score. AutoGrade remains consistent by using the damage descriptions." (https://www.naaa.com/References/reference_materials/AutoGrade_Information_Document.pdf) [source]
**B1. Under the NAAA Arbitration Guidelines effective June 1, 2026, the repair cost used for arbitration thresholds is determined by the auction at local wholesale rates.** Section V.2.d reads: "The auction will only inspect the mechanical defect(s) on the initial arbitration claim. For purposes of the dollar thresholds for arbitration, the repair cost will be determined by the auction based upon the local auction/wholesale cost to repair the defect(s)." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) [source]
**F8. The five damage classes that automated estimators handle well are precisely the classes AutoGrade files under Minor and Moderate.** Dents, scratches, cracks, broken lamps, and glass shatter map to AutoGrade's Moderate (glass/windshield, body damage/repair) and Minor (cosmetic body damage, bumpers/partial) bands. (https://www.naaa.com/References/reference_materials/AutoGrade_Information_Document.pdf) The Major band — structural, mechanical/engine, fire/flood/biohazard, non-drivable — is where the cost variance and the arbitration exposure live, and is exactly what image-based estimation d [source]
**U2. Whose repair cost estimate is authoritative is settled contractually but contested economically.** The guidelines assign the number to the auction at local wholesale rates (B1). A buyer's realistic cost is a retail or in-house shop cost, which can exceed local wholesale. The guidelines do not address this gap, and no source located reports how often auction-determined costs fall below buyer-obtained estimates or by how much. This is the single most consequential unmeasured quantity in the concept. [source]
**D6 — Provenance of the salvage-lot repair-cost figure is undocumented.** No source found states, on the record, how sellers derive the "Repair Cost" they submit to Copart (carrier estimate, internal formula, or otherwise). Claim 7 is an inference and should be treated as an open question, not a finding. [source]
Terminology note: this report uses "repair cost estimate" for a dollar figure, and "condition grade" or "condition score" for the 0–5 scale. The distinction matters — see Claim 3. [source]
**Claim 6a — ACV Auctions made the parallel move in February 2022, acquiring Monk SAS.** ACV announced the acquisition of Monk SAS, an AI-powered imaging and vehicle inspection platform, on February 28, 2022, positioning it alongside ACV's existing condition report features Virtual Lift® and Audio Motor Profile (AMP®). Automotive News reported the price at approximately $19 million. Sources: https://investors.acvauto.com/news-events/press-releases/detail/52/acv-acquires-monk-sas-a-leading-ai-powered-imaging-and-vehicle-inspection-platform · https://www.autonews.com/dealers/acv-acquires-vehicle [source]
Primary — standards and policy - NAAA Auction Condition Report Position Statement, updated June 1, 2026 — https://site.manheim.com/en/marketplace-policies/us-policies/naaa-auction-condition-report-position-statement.html - NAAA Arbitration Guidelines, In-Lane and Online, effective June 1, 2026 — https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf - NAAA Arbitration Guidelines Appendix I, effective June 1, 2026 — https://naaa.com/wp-content/uploads/2026/05/For_Printing_Arbitration_Guidelines_Appendix_I_June_1_2026.pdf - NAAA Vehicle Condition Grading [source]
11. The condition grade and the repair cost estimate are deliberately separate quantities computed from the same damage lines. NAAA's AutoGrade document lists "Estimated Repair Costs" in its table of items that do **not** impact AutoGrade, alongside mileage, model year, make, trim, color, options, and announcements. (https://naaa.com/wp-content/uploads/2022/11/Autograde_Std_Vehicle_Condition_Grade_5_15_2018.pdf) [source]
12. The stated reason for that exclusion is rate variance. The same document's Q&A asks "Does AutoGrade™ consider the cost of repairs?" and answers: "No – Repair labor rates can vary from shop to shop, skewing the grade. AutoGrade™ remains consistent by using the damage descriptions." It adds that "Grade is not determined by the price of the repairs." (https://naaa.com/wp-content/uploads/2022/11/Autograde_Std_Vehicle_Condition_Grade_5_15_2018.pdf) [source]
18. The cost of resolving a disputed structural measurement is allocated by its outcome, which prices the act of estimating itself. NAAA: "If the measuring facility determines that the vehicle is within the UVMS, the buyer of the vehicle will be responsible for the charges paid to the facility. Likewise, if the measuring facility determines that the vehicle is not within the UVMS, the seller will be responsible for the charges." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf, §VIII.4.e) [source]
23. Database output is not interchangeable across providers, so a repair cost figure is only meaningful relative to the system that produced it. The same source notes the three U.S. providers (Audatex, Mitchell, CCC/MOTOR) all apply manufacturer data as "a guide to establish flat rate times," and that "the way the information is applied or calculated may differ between the information providers." (https://degweb.org/know-your-estimating-system/; https://scrs.com/estimating-tip-ccc-mitchell-and-audatex-sectioning-operations/) [source]
26. The estimate's marginal information value collapses for a physically present buyer. Tadelis and Zettelmeyer report that "auction bidders are quite experienced in assessing a vehicle's condition," that "A relatively quick up-close visual inspection can identify to a large degree whether the condition score ought to be low, high, or somewhere in between," and conclude that "conditional on a bidder showing up at a vehicle's auction, the information revealed by the SCR is not very useful." (https://faculty.haas.berkeley.edu/stadelis/autoauction_073114.pdf, p. 5) This is the strongest disconfir [source]
1. **Manheim states two different per-defect floors for the same product.** The Digital Buyer Protection policy page says "Each cosmetic or structural defect must meet a minimum estimated wholesale repair cost of at least $400" (https://site.manheim.com/en/marketplace-policies/us-policies/digital-buyer-protection.html), while the Manheim Buyer Arbitration Guide describes the same product's defects as needing to be "singularly $200 or more and reach a cumulative total of $800" (https://site.manheim.com/tutorials/manheim-buyer-arbitration-guide). The $800 aggregate is consistent across both. Not [source]
**B8. Grades themselves are not arbitrable — only undisclosed damage is.** Section V.4.h: "Auction and seller make no representations or guarantees of vehicle grades or other types of scoring systems. Buyers may only arbitrate a vehicle based upon undisclosed damage or defects that were present at the time of the sale." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) A buyer who priced recon off the grade and was wrong has no remedy from the grade being wrong. [source]
**C3. The cost of the measurement itself is allocated by its outcome.** Section VIII.4.e: if the measuring facility finds the vehicle within UVMS, the buyer pays the facility charges; if it finds the vehicle outside UVMS, the seller pays. (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) Verifying an estimate therefore carries its own expected cost, which belongs in the decision to verify. [source]
**E2. The estimate is expected to move upward materially as information improves.** A worked example tracks one vehicle from $1,200 initially, to $1,850 after photos revealed windshield and tire damage, to $2,450 after technicians found additional repairs post-purchase — roughly a doubling from first estimate to final. (https://getcarbly.com/blog/your-recon-allowance-is-not-a-fixed-number/) [source]
**F9. Commercial photo-based estimators position their output as a range, not a figure.** CCC's consumer-facing tool gives "a ballpark range of what the repair may cost" from uploaded photos, and with seven photos "can identify damaged panels, suggest preliminary repair lines, and prepopulate an estimate" — prepopulating an estimator, not replacing it. (https://www.cccis.com/news-and-insights/posts/how-to-put-the-ai-in-repair) [source]
**U3. Whether automated estimation is close to production-grade is genuinely disputed by the evidence.** Vendor material describes AI estimating as deployed and resolving claims in minutes (https://www.cccis.com/news-and-insights/posts/how-to-put-the-ai-in-repair), while the peer-reviewed work that reports strong numbers does so on synthetic cost labels in a single currency with no external validation (F6, F7), and the industry's own supplement data shows human experts missing half of all calibrations in person (F2). These are not straightforwardly reconcilable. The most defensible reading is [source]
6. **The "Est. Retail Value" / ACV field carries the same disclaimer** and is the pre-damage value, not the as-is value. Using repair-cost ÷ ACV as a damage-severity ratio therefore divides one disclaimed seller number by another. **[Med]** (same URL) [source]
13. The grade instead runs on an ordinal severity scheme. AutoGrade sorts damage line items into Major, Moderate, and Minor grade reductions, and notes that "Damage Size / Severity / Number of Occurrences per damage line item may have a graduated impact on total grade." Manheim's patent announcement describes the underlying algorithm as "applying an objective point scale to the damages." (https://naaa.com/wp-content/uploads/2022/11/Autograde_Std_Vehicle_Condition_Grade_5_15_2018.pdf; https://press.manheim.com/2012-09-26-Manheim-Awarded-Patent-For-Computer-Assisted-Condition-Reporting-And-Autog [source]
- ~/.global-ai-hub/research-runs/used-car-auctions-20260919/auction-vehicle-repair-cost-estimation/synthesis.md — new: merged synthesis of four reports [source]
3. **Named omissions from a generic CR** include mechanical issues with the engine, transmission and drivetrain; non-visible structural issues; and open safety recalls. A cost model built on CR line items has a systematic negative bias for exactly these categories. **[High]** (same URL as claim 1) [source]
4. **AI cost-estimation claims are unverified.** Vendor tools advertise photo-to-repair-cost ranges, while the peer-reviewed literature stops at detection and flags severity classification as subjective (Claim 8). No independent benchmark of vendor cost-estimate accuracy against actual repair invoices was located. [source]
**Met.** Nine independent hosts were used, including four primary sources (two NAAA standards documents, one Manheim/NAAA product specification, one peer-reviewed AER field experiment) and two dated corporate primary announcements (Cox Automotive press release, ACV investor relations). A disconfirming source was actively sought and found in two forms: the NAAA position statement disclaiming repair-cost estimation (Claim 1), and the academic/industry evidence that automated detection does not yet support reliable costing (Claims 8, 8a). [source]
Secondary — trade press and vendor history - Mitchell, Our History — https://www.mitchell.com/about/our-history - Mitchell 1 History (vendor self-reported, 1958 labor-times date) — https://mitchell1.com/history/ - Solera Canada / Audatex History (1966 founding) — https://www.audatex.ca/about/history/ - Database Enhancement Gateway, Know Your Estimating System — https://degweb.org/know-your-estimating-system/ - Repairer Driven News on the CCC 2023 Crash Course report (calibration supplement rate) — https://www.repairerdrivennews.com/2023/03/14/ccc-2023-crash-course-report-explores-ai-effects-on [source]
2. The estimate's target state is cosmetic, not mechanical. The same source states the figure prices the vehicle to a state where "exterior and interior damage are no longer noticeable," and that the underlying inspections "do not include the mechanical condition of the car, except that the inspecting technician documents unusual engine sounds." (https://faculty.haas.berkeley.edu/stadelis/autoauction_073114.pdf, p. 5) [source]
Out of scope and deliberately not researched: valuation/pricing models (MMR and the like), auction bidding mechanics, title and salvage branding, retail pricing strategy, and collision-insurance claims practice except where it is used as a calibration benchmark for estimate accuracy. [source]
**C2. Visual evidence outranks instrumented measurement in that adjudication.** Section VIII.4.f states flatly: "Visual evidence supersedes any/all mechanical or electronic measurements." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) This is a live constraint on any sensor-based or AI-based structural cost estimator: a more precise measurement does not win the dispute. [source]
**E1. The working practice is a default allowance keyed to condition-report score, then revised as evidence arrives.** Dealers "set default recon costs by condition-report score" and roll those defaults into vehicle-level profit calculations, treating them as "a consistent starting point." (https://getcarbly.com/blog/your-recon-allowance-is-not-a-fixed-number/) [source]
**G4. Do not spend reconditioning money on a vehicle with an open or likely arbitration claim.** IV.2 makes that spend unrecoverable (B6), and the vehicle must be returned "in the same or similar condition, consistent with its marketable condition at the time of sale." (https://naaa.com/wp-content/uploads/2026/05/NAAA_Arbitration_Guidelines_June_1_2026_UPDATED.pdf) [source]
**U5. Whether reconditioning cost benchmarks are current.** The $500–$1,500 range (E5) is a vendor-blog figure without a stated methodology or vintage. Meanwhile CCC's collision severity is rising (F5). Whether wholesale recon costs have tracked that rise is not established by any source located. [source]
**One source could not be retrieved.** The systematic literature review "Vehicle Damage Detection Using Artificial Intelligence" (Hasan, 2025, *WIREs Data Mining and Knowledge Discovery*, DOI 10.1002/widm.70027) returned HTTP 403 and its contents are therefore **not** cited anywhere in the claims above. Its existence is noted only as a pointer for follow-up. [source]
**Retrieved but unusable (not cited in any claim):** 18. *WIREs Data Mining and Knowledge Discovery* — Vehicle Damage Detection Using Artificial Intelligence: A Systematic Literature Review (Hasan, 2025), DOI 10.1002/widm.70027 — HTTP 403, contents not read — https://wires.onlinelibrary.wiley.com/doi/10.1002/widm.70027 [source]

Related concepts

Cost — is a part of Auction Vehicle Repair Cost Estimation
Vehicle — is a part of Auction Vehicle Repair Cost Estimation
Repair — is a part of Auction Vehicle Repair Cost Estimation
Auction — is a part of Auction Vehicle Repair Cost Estimation
Estimation — is a part of Auction Vehicle Repair Cost Estimation

Children

No children recorded.

← the whole tree · 3D view· how to read this page