What the Allstate McKinsey and Colossus Records Actually Say

Short answer: Allstate's fight to keep its McKinsey claims documents sealed is adjudicated and documented. What the documents show is not. The 2010 Colossus settlement found a governance defect in how a claims algorithm was tuned and recorded no finding of underpayment.

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

Allstate spent years fighting to keep a set of consultant documents out of public view. That fight, not the contents of the documents, is the part of this story the public record proves.

The version in wide circulation runs like this: a consultant taught Allstate to deny, delay and refuse to pay; a court caught the company hiding the evidence; regulators later confirmed software was used to underpay claims. Every load-bearing element of that is either contradicted by the operative documents or unverifiable at source. What the record supports is narrower and more useful to a repairer: a company that took adjudicated sanctions rather than hand over its claims-redesign records, and an examination that found a governance defect in how a claims algorithm was calibrated while expressly declining to find underpayment.

Two things a court actually decided

On 17 January 2008, effective at 1 p.m., Florida Insurance Commissioner Kevin McCarty suspended the certificates of authority of ten Allstate and Encompass companies, barring them from writing new business in all lines in the state. Renewals and existing policies were untouched. McCarty was blunt: "In view of Allstate's ongoing, blatant disregard of our subpoenas, I have little choice but to take an action that will send a clear message about how seriously I am taking this issue." (Insurance Journal, 2008)

Allstate appealed and obtained a stay. It lost. On 4 April 2008, in Case No. 1D08-0275, Florida's First District Court of Appeal affirmed: "Because the IFO facially complies with the requirements of section 120.60(6), Florida Statutes, it is AFFIRMED and the stay is lifted." (First DCA order)

The scope of that proceeding is routinely overstated. The subpoenas served in October 2007 concerned the companies' relationships with rating agencies, modelling firms and trade groups, and how those might have influenced rate increases requested that September (Insurance Journal). The McKinsey material sat inside that inquiry rather than defining it, but it was squarely in it: the appellate order records counsel offering the McKinsey documents only "subject to the appropriate protections" and "privileges," and that "Allstate did not produce witnesses to respond to questions regarding Allstate's claims handling practices as contained in the McKinsey report... despite being requested to do so" (First DCA order).

The same order contains the most useful corroborated fact in the affair: "At the time the IFO was entered, Allstate had incurred approximately $2.4 million in court fines for failure to comply with court-ordered production." One state's appellate court recorded, in an affirmed order, that another state's had already fined Allstate millions for refusing to produce.

Neither the suspension order nor the appellate decision affirming it contains any finding about how Allstate handled claims. Both are about compliance with subpoenas. The distinction holds for the whole of this history: fighting disclosure and being sanctioned for it is one proposition, and having done the thing the documents were sought to prove is another. Only the first is adjudicated.

The sanction everybody cites was reversed

The most-cited McKinsey sanction is a New Mexico default judgment. It did not survive.

The trial court had ordered Allstate to produce the documents without a protective order and, when Allstate refused, entered default judgment against it on liability. In Pincheira v. Allstate Insurance Co., 2008-NMSC-049, the New Mexico Supreme Court held otherwise: "Because Defendant met its initial burden of making a good faith claim of a trade secret, the trial court's denial of Defendant's request for a protective order and evidentiary hearing was an abuse of discretion." The disposition is unambiguous: "We remand to the trial court to reverse its entry of default judgment and to continue this case on its merits." (Pincheira v. Allstate, N.M. Sup. Ct. No. 30,490; Justia)

Two things that opinion does not do. It does not hold the documents were trade secrets, only that Allstate had said enough to earn a hearing. And it says nothing about what is in them. The case was about discoverability, not content.

It also records how the dispute ended: "We agree that Defendant's public disclosure of the McKinsey documents has rendered the request for a protective order moot" (Pincheira). Allstate put the documents into the public domain itself. Anyone writing that a court stripped it of its trade-secret claim has the history backwards.

What the Colossus regulators actually found

Colossus is the other half of the legend, and its operative document is unusually clear. The 2010 Multi-State Market Conduct Regulatory Settlement Agreement, filed as Oregon administrative order 10-09-021-RSA, describes the software in its own recitals: "The Colossus software program is a rule based program with approximately six hundred (600) injury profiles and is a tool Allstate adjusters use in evaluating general damages for certain bodily injury claims," used on "approximately fifty percent (50%) of all automobile bodily injury claims." (Colossus settlement agreement)

Note the scope. Colossus values general damages on bodily injury claims. It does not price parts, labour or refinish, and is not a collision estimating system. Treating a Colossus finding as one about repair estimates is the commonest error here. Now the findings, verbatim. On underpayment: "Based on review of the materials, information and documents produced in the course of the Examination, the Lead Regulators did not identify institutional issues involving underpayment of claims." On whether adjusters were made to settle inside the software's number: "Allstate resolves over ninety percent (90%) of the claims on which the Colossus software program is used within the Evaluation Range. There is no evidence that Allstate established a policy or rule requiring its claims adjusters and/or Evaluation Consultants to settle claims within a range of the value recommended by the Colossus software program." Both sentences belong in any honest account. And on what was actually examined: "Individual claims files were not reviewed. Consequently, there is no finding regarding the payment of any particular bodily injury claim." (Colossus settlement agreement)

The defect regulators did find was governance of "tuning," which the agreement defines as "the process whereby recent settlement data is mapped onto trauma severity points": "The Examination determined that there should be enhanced management oversight to ensure adherence to established criteria for selection of claims to use in the tuning process and a uniform methodology for determining the number of claims to be used in each tuning region. However, the Examination found no evidence of improprieties with respect to any particular claim selection or tuning." (Colossus settlement agreement)

The remedies match the defect. Allstate undertook to adopt the sampling and management-oversight practices set out in the agreement's exhibits, to submit the evaluation process to internal audit, to notify claimants that Colossus may be used, to document its reasons for declining any CSC update to the software, and not to "establish a policy or rule requiring its adjusters and/or Evaluation Consultants to settle claims based solely on the value recommended by the Colossus software program." The lead regulators were to run compliance reviews annually "for a period of four (4) years." The money was a $10,000,000 fund for the signatory regulators to spend on "developing and training examination personnel in processes designed to review and monitor the insurance industry's use of software technology tools in the claims handling process"; none of it went to claimants. Forty-five states signed, and the agreement ran "until December 31, 2015." It carries a non-admission that belongs in the same breath as the money: Allstate "does not admit, deny or concede any actual or potential fault, wrongdoing or liability in connection with any facts or claims that have been or could have been alleged against it." (Colossus settlement agreement; Insurance Journal, 2010)

New York Superintendent James Wrynn put it plainly: "It is important to note that we found no systemic underpayment of bodily injury claims." Allstate quoted the settlement's own phrase back, that the examination "did not identify institutional issues involving underpayment of claims" (Insurance Journal).

What survives is still substantial: regulators found a widely used claims algorithm being recalibrated region by region without uniform, supervised adherence to the criteria governing which claims fed the recalibration, and imposed sampling standards, management oversight and an audit regime. That is an early precedent for regulating algorithmic claims handling, and it is not a finding that anybody was underpaid. We are not saying Allstate underpaid bodily injury claims through Colossus: the lead regulators wrote that they did not identify institutional issues involving underpayment, and they also wrote that individual claims files were not reviewed, so the examination is not evidence in either direction about any particular claim.

What was in the documents, and where they are now

The Missouri side of the fight is corroborated by the primary record but not supplied by it. An Associated Press report of 11 July 2008, reprinted on a plaintiff firm's blog, states that Jackson County Judge Michael Manners held Allstate in contempt in September 2007 and "began fining it $25,000 per day," that "the Missouri Supreme Court in November ordered Allstate to turn over the documents, leading the company to disclose more than 120,000 pages of records," and that Allstate's reluctance "led to more than $7 million in fines" before a settlement whose approval hearing was set for 21 July 2008 (Associated Press, via Robin Frazer Clark P.C.). That is a wire report read on a secondary host, and the Missouri docket was not obtainable. Primary is the Florida court's recital of roughly $2.4 million in fines by January 2008.

In April 2008 Allstate posted approximately 150,000 pages publicly. That same week Consumer Watchdog recorded that "Allstate filed a list of documents it objects to submitting to the OIR totaling 196 pages, arguing they are privileged and involve trade secrets," and quoted Ed Domansky of Florida's Office of Insurance Regulation on whether the published set matched what the regulator had received: "We don't really know if it's the same." (Consumer Watchdog, 2008)

Then the twist that almost never gets reported. A consumer advocate with no interest in defending Allstate read the released pages and found they did not contain what everyone expected. Amy Bach, executive director of United Policyholders: "There is no blatant smoking gun." She added: "Nothing stands on its own in those 150,000 pages. It's like a giant puzzle and you have to put the pieces together and connect the dots." The complaint in the same report was about an absence, and it belongs to two other advocates: Robert Hunter of the Consumer Federation of America and Birny Birnbaum of the Center for Economic Justice, who said Allstate should also share "e-mails and memos that show which McKinsey recommendations the insurer had implemented." Florida's subpoena had asked for exactly that (United Policyholders). Chip Merlin, a policyholder-side coverage lawyer, had already read the documents under a secrecy order that barred him from describing them, and reached a similar place from the other direction. "Documents which may indicate cultural or systematic issues are secondary," he wrote, because "establishing bad faith liability through case specific breaches of good faith is the most important aspect of a bad faith trial" (Merlin, 2007).

We publish no quotation from the slides themselves. Allstate's document-release page returned a connection failure on 1 September 2026, and no live public repository of the released set could be read on 2026-09-01. Every slide quotation in circulation travels through secondary retelling and cannot be checked at source. Documents released to end an argument about secrecy are, eighteen years later, effectively unavailable. The slides could be harmless. They could be damning. Everyone arguing either way is working from a retelling, and so would we be. Neither position can be checked against a page anyone can now open.

Two phrases should be retired. "Deny, delay, don't pay" and "the three Ds" are plaintiff-side characterisations, not language quoted from the documents. "Delay, Deny, Defend" is the title of Jay Feinman's book about the industry generally, not an Allstate or McKinsey document.

The insurer that owned the body shops

The most striking fact in Allstate's record has nothing to do with McKinsey. In May 2001 The Allstate Corporation acquired Sterling Collision Centers Inc., then 39 auto collision repair stores in seven states and nine metropolitan locations. Terms were not disclosed. The structure was deliberate and stated flatly in the announcement: "Sterling will operate as a separate subsidiary of Allstate Non-insurance Holdings, Inc. and will not be owned by the Allstate Insurance Company." George Ruebenson, Allstate's vice president for claims, became chairman of Sterling's board of directors and framed the purchase as a way to "create a differentiated customer experience" and increase "efficiencies and delivery of value" to policyholders. (Insurance Journal, 2001)

For thirteen years a top-five US auto insurer sat on both sides of the repair transaction, writing the estimate and owning a chain that did the work. Housing the shops in a non-insurance holding company rather than the carrier is exactly the arrangement anti-steering and referral rules are written to think about.

In April 2014 Allstate agreed to sell Sterling to Service King Collision Repair Centers. By then it ran 62 facilities in 16 states with about 1,200 employees. The price was not disclosed, Allstate expected a small after-tax gain, and Service King had already repaired Allstate customers' vehicles for more than 30 years as part of the Good Hands Repair Network (Carrier Management, 2014).

No public rationale for the exit was found, and we will not invent one. We are not claiming to know why Allstate sold the chain. Thirteen years of ownership took the chain from 39 shops to 62, and the exit is complete: no Sterling entity appears in Allstate's FY2025 Exhibit 21 subsidiary list, though Allstate Non-Insurance Holdings, Inc. is still there (SEC EDGAR, Exhibit 21, FY2025 Form 10-K). The holding company outlived the shops.

What is documented about Allstate collision claims now

The most collision-relevant regulatory document here is recent, primary and almost unreported. Pennsylvania's Insurance Department examined Allstate Fire and Casualty Insurance Company for calendar year 2023 and issued its report and consent order on 4 November 2025. In auto collision, examiners sampled 75 claim files, cited 31 violations and found violations in 21 files: an error ratio of 28 percent, the worst of any auto category examined. Comprehensive was 15 percent, property damage 11, first-party medical referred to a peer review organisation 20, total loss 1, and first-party medical not referred to a peer review organisation 0 (Pennsylvania exam report).

The collision findings are about appraisal quality specifically. Within that category, 21 violations were cited under 31 Pa. Code 62.3, requiring that "An appraisal shall meet all applicable standards per statute," and 10 under 63 P.S. 861(b), requiring that an appraisal contain "the number of the appraiser's license and the proper identification number of the vehicle being inspected." A separate total-loss violation concerned sales tax, for which restitution was provided. The Department made eight recommendations. The penalty was $5,000. (Pennsylvania market conduct report and consent order)

More than a quarter of sampled collision files, in the coverage that pays body shops, carried an appraisal defect or a missing appraiser licence number. The penalty was five thousand dollars. Both halves are findings.

Vermont's regulator reached the licensing question directly. In Docket No. 21-024-I, dated 8 December 2021, the Department of Financial Regulation imposed a $225,000 penalty on four Allstate entities. The Department identified violations including "Failure to attempt, in good faith, to effectuate prompt, fair, and equitable settlements of claims in which liability has become reasonably clear," and recorded this: "The Department identified numerous claims that were settled without having been reviewed by a Vermont licensed adjuster." Allstate agreed to pay restitution with statutory interest to third-party claimants denied on comparative-negligence assignments later found faulty, and conceded nothing: respondents "do not admit the existence of violations." (Vermont DFR, Docket No. 21-024-I)

Read Vermont and Pennsylvania together and one thread appears: both are about who is licensed to produce the number a shop is paid from, and whether the file shows it. It sharpened in 2017, when Allstate began closing its auto drive-in estimating centres for the QuickFoto Claim app; by that August roughly half of all driveable vehicles were inspected that way (Claims Journal, 2017).

On the network side, Allstate's 11 January 2021 directive told Good Hands Repair Network shops to "evaluate the need to perform vehicle scanning operations on a case by case basis when it is necessary to properly repair customer's vehicles," weighing vehicle age, damage, ADAS equipment, airbag deployment and "if indicated, by a vehicle specific OEM repair procedure." The reimbursement is explicit: "Allstate will pay .5 hours mechanical rate for a Pre-Scan when necessary and .5 mechanical rate for a Post-Scan," with a remote alternative "at a charge of up to $125.00 and .5 hours body rate." Work the shop could not do itself was left to negotiation: "Operations not able to be completed by the repair facility, such as diagnostic services, system or equipment calibrations, reprogramming, initializations, and other services should be charged at a competitive cost." (Repairer Driven News, 2021)

In October 2025 that last item changed. Allstate announced an Opus IVS agreement offering network shops "a flat-rate five-tier calibration pricing structure and reimbursement of the $3 ADAS MAP report fee," integrated with CCC estimating. The trade report notes what the announcement did not say: "The email does not explain the flat-rate five-tier calibration pricing structure, nor does it provide the prices." Whether participation is voluntary is not stated either (Repairer Driven News, 2025).

Calibration reimbursement thus moved from negotiated competitive cost to an insurer-set tier structure whose numbers are unpublished. Nobody outside the network can read them.

The arithmetic behind the pressure

None of this requires a conspiracy to explain the incentive. For FY2025 Allstate reported a Property-Liability combined ratio of 85.2 on $57.682 billion of premiums earned, meaning roughly 14.8 cents of every Property-Liability premium dollar was kept as underwriting profit before any investment return (Allstate Q4 and full-year 2025 earnings release). With the expense side broadly fixed, the swing in that ratio is loss ratio, and loss ratio in auto physical damage is substantially what a repair costs. Allstate's own explanation of $719 million of fourth-quarter prior-year non-catastrophe reserve reestimates names the coverages: "favorable severity development in personal auto injury and physical damage coverages" (Allstate Q4 and full-year 2025 earnings release; Repairer Driven News, 2026). Prior-year claims settled for less than reserved, in the two coverages at the centre of this history. That is a disclosure, not a wrongdoing, and the clearest published statement of where pressure on a repair estimate comes from.

One line of litigation cuts the other way. Body shops brought federal antitrust claims against roughly forty carriers, Allstate entities among the defendants, consolidated as In re Auto Body Shop Antitrust Litigation, MDL No. 2557, before Judge Gregory Presnell in the Middle District of Florida. He dismissed the price-fixing and boycott claims at the pleading stage with leave to replead, applying the Supreme Court's rule that "lawful parallel conduct fails to bespeak unlawful agreement" (National Law Review). The Eleventh Circuit, sitting en banc, "affirmed the dismissal of the federal antitrust claims and all the state law claims, except the tortious interference claims"; when the Judicial Panel on Multidistrict Litigation remanded the five surviving actions in December 2020, "No discovery has commenced" (JPML remand order, MDL No. 2557). We do not claim the conduct those shops alleged occurred. No court has ruled on whether it did.

What we could not establish

We cut the following rather than publish it.

The McKinsey slide quotations. This is the central wall of the subject, and it is part of the story rather than a footnote to it. Allstate's own document-release page, on the host media.allstate.com, did not answer: the connection timed out, which is not evidence the page is gone. No other live public repository of the roughly 150,000 released pages could be read on 2026-09-01 either. Every slogan and slide title attributed to the set, including the handful that circulate most widely, reaches the public only through secondary retelling and cannot be checked against the page it is said to come from. We reproduce none of them. We do not claim to know what is on them. Documents published in April 2008 to end an argument about secrecy are, eighteen years later, effectively unavailable.

"Deceptive Practices, Inc." and "Three-D" as case names. No litigant by either name was found in any Allstate McKinsey matter. They appear to conflate deceptive-trade-practices causes of action with the "three Ds" shorthand, and should not be cited as cases.

The Missouri case caption. The contempt is corroborated in a primary appellate order, but the Missouri record was not obtainable at any live source and the case name is not established. The $25,000 per day rate and the figure of more than $7 million in fines rest on a single Associated Press report.

Figures that would not hold still. Two reputable sources give different days in 2008 for the Pincheira opinion, so we cite it by official citation and year. Repeated readings of the Pennsylvania report returned different aggregate violation totals and different aggregate counts under 31 Pa. Code 62.3; the per-category figures, sample sizes, error ratios, penalty and recommendation count were stable across every reading and are published, the aggregates are not. Filings and secondary summaries also disagree on Allstate's FY2025 underwriting income and subsidiary count, so only figures read directly from the filings appear above.

Vermont's signed order. The underlying PDF is an image-only scan with no text layer, so it cannot be string matched. The quotations come from the Department's own published order page, which serves the full text as readable HTML, not from the executed document.

Why Allstate exited collision repair ownership. No Allstate statement of rationale for the 2014 Sterling sale was found, and no regulator's examination or opinion on Allstate owning repair facilities while directing claims to them was located. Regulatory exposure, margin and conflict of interest are all plausible; none is evidenced. We publish no motive.

Good Hands Repair Network terms. No shop count, current or historical, was found, and no Allstate parts-usage directive or parts-related class action specific to Allstate was located. Whether the Opus IVS calibration tiers are a floor or a ceiling, and whether participation is mandatory, is not stated anywhere published.

Loose ends. Whether the 196 withheld pages were ever produced, and whether Florida obtained the implementation records consumer advocates said were missing, is unknown. So is whether the Colossus compliance reviews, required annually for four years, found compliance, and whether any tuning control survived the agreement's expiry.

Advocacy figures on Allstate's payout ratios and pre-tax income before and after 1995, and asserted Maryland and Texas penalties circulating in a 2008 trial-lawyer report. None was verified against a primary record, and none is published here.

Corrections

This page is new, so nothing on it has yet been corrected after publication. What follows are the corrections made to its own draft before it went up, logged here because a page that shows only its wins is not evidence of anything.

2026-09-01, a cut list that republished what it disclaimed. The draft's own list of unpublishable material stated that it quoted no slide, then printed a verbatim slide slogan and a purported slide title one sentence later. Both strings are gone, and the page now reproduces no text from the McKinsey slides anywhere, in the body or in the cut list. The draft had also written that it could not settle whether one of those phrases was a slide title or a commentator's label; the wire report it cites does describe it as a slide title, so the hedge was wrong as well. The phrase was cut rather than asserted, because it still cannot be checked against the slide. A disclaimer that republishes the thing it disclaims is not a disclaimer.

2026-09-01, the antitrust paragraph. The draft had Judge Presnell holding that lawful parallel conduct fails to bespeak unlawful agreement as if it were his own rule. It is the Supreme Court's, from Bell Atlantic v. Twombly, and he applied it. The draft also presented the dismissal as final, omitting that it came with leave to replead, and rested Allstate's presence in the case on a National Law Review piece that does not mention Allstate at all. The JPML remand order for MDL No. 2557, which names the Allstate defendants, was added; so were the Eleventh Circuit's en banc disposition and the December 2020 remand at which "No discovery has commenced". The paragraph now closes by saying no court has ruled on whether the conduct alleged occurred.

2026-09-01, the Colossus defect, overstated. The draft said regulators found the algorithm being recalibrated without documented control over which claims fed the recalibration. The settlement records that established criteria existed; what the examination found was that adherence to them needed enhanced management oversight, and that the methodology for deciding how many claims to use in each tuning region was not uniform. The sentence now reads "without uniform, supervised adherence to the criteria governing which claims fed the recalibration". The defect is narrower than the draft made it, and it is the defect the document actually names.

2026-09-01, the missing implementation records, misattributed. The draft credited United Policyholders with the complaint that Allstate should also release the e-mails and memos showing which McKinsey recommendations it had implemented. The cited page attributes that to Robert Hunter of the Consumer Federation of America and Birny Birnbaum of the Center for Economic Justice. It is now attributed to them, and the cut list was realigned to match.

2026-09-01, the only clean Pennsylvania category, omitted. The draft listed every auto category in the 2025 examination except the one with an error ratio of 0, first-party medical not referred to a peer review organisation. Dropping the single category the company passed cleanly is a cherry-pick however it is meant. It now sits in the list with the others.

2026-09-01, the Florida orders, left unguarded. The draft recited the suspension, the appellate affirmance and the approximately $2.4 million in court fines, then moved on, at exactly the point where a reader concludes the orders found something about claims handling. An exact-string check of the appellate order found no such finding in it. The page now says so in the same section: both orders are about compliance with subpoenas, and only the disclosure fight is adjudicated.

Related

• Why claims against insurers almost never succeed
• The rate loop: how the prevailing labour rate is set
• AI is grading your auto claim, and almost no law requires a human to check it

Sources

• Multi-State Market Conduct Regulatory Settlement Agreement, Oregon order 10-09-021-RSA, the operative 2010 Colossus settlement, and the source of every quoted recital here about what the software does, the tuning defect, the remedies, the $10,000,000 fund, the four-year compliance reviews and the non-admission. Read on 2026-09-01.
• Allstate v. Florida OIR, First District Court of Appeal, Case No. 1D08-0275, the 4 April 2008 order affirming the immediate final order and lifting the stay, and the primary record of the approximately $2.4 million in court fines and of the witnesses not produced. Read on 2026-09-01.
• Pincheira v. Allstate Insurance Co., New Mexico Supreme Court, on FindLaw, the 2008 opinion reversing the default judgment, supporting the trade-secret holding, the mootness passage and the fact that the opinion describes no contents. Read on 2026-09-01.
• Pincheira v. Allstate Insurance Co., on Justia, the second host used to check the same opinion string for string, and the source of the conflicting decision date noted above. Read on 2026-09-01.
• Insurance Journal, 18 January 2008, trade report of the Florida suspension, its effective hour and scope, Commissioner McCarty's statement and the subject matter of the October 2007 subpoenas. Read on 2026-09-01.
• Insurance Journal, 19 October 2010, trade report of the Colossus settlement, the source for the count of signatory states and for Superintendent Wrynn's statement. Read on 2026-09-01.
• Insurance Journal, 15 May 2001, the Sterling Collision Centers acquisition announcement, its store and state counts, the holding-company structure and George Ruebenson's board chairmanship. Read on 2026-09-01.
• Consumer Watchdog, April 2008, contemporaneous account of the document release, the 196-page objection list and Ed Domansky's comment for the Florida Office of Insurance Regulation. Read on 2026-09-01.
• United Policyholders, the source for Amy Bach's assessment of the released pages and for the Hunter and Birnbaum complaint about missing implementation records. Read on 2026-09-01.
• Merlin, October 2007, a policyholder-side lawyer's account of having read the documents under a secrecy order, and his view of their evidentiary weight. Read on 2026-09-01.
• Associated Press, 11 July 2008, reprinted by Robin Frazer Clark P.C., the wire report and the only source for the Missouri contempt, the $25,000 per day rate, the 120,000 pages and the more than $7 million in fines. Read on 2026-09-01.
• Carrier Management, 15 April 2014, the Sterling sale to Service King, the facility, state and employee counts, and the absence of any stated rationale. Read on 2026-09-01.
• SEC EDGAR, Allstate FY2025 Form 10-K, Exhibit 21, the subsidiary list checked for Sterling entities and for Allstate Non-Insurance Holdings, Inc. Read on 2026-09-01.
• Pennsylvania Insurance Department market conduct report and consent order, 4 November 2025, the examination of Allstate Fire and Casualty Insurance Company for 2023, and the source of every per-category sample size, error ratio, cited provision, recommendation count and the $5,000 penalty. Read on 2026-09-01.
• Vermont Department of Financial Regulation, Docket No. 21-024-I, the 8 December 2021 stipulation and consent order, the $225,000 penalty, the licensed-adjuster finding and the respondents' non-admission. Read on 2026-09-01.
• Claims Journal, 7 August 2017, the drive-in estimating centre closures and the share of driveable vehicles inspected through the QuickFoto Claim app. Read on 2026-09-01.
• Repairer Driven News, 22 February 2021, the 11 January 2021 scanning directive to Good Hands Repair Network shops and its reimbursement terms. Read on 2026-09-01.
• Repairer Driven News, 28 October 2025, the Opus IVS calibration agreement, the $3 ADAS MAP report fee and the fact that the prices were not published. Read on 2026-09-01.
• Allstate fourth quarter and full year 2025 earnings release, the primary filing for the 85.2 combined ratio, the $57.682 billion of premiums earned and the $719 million of prior-year non-catastrophe reserve reestimates. Read on 2026-09-01.
• Repairer Driven News, 12 February 2026, the trade report of the same results, cited alongside the primary release rather than in place of it. Read on 2026-09-01.
• National Law Review, the account of Judge Presnell's pleading-stage dismissal in MDL No. 2557 and of the Supreme Court rule he applied. It does not name Allstate. Read on 2026-09-01.
• JPML remand order, MDL No. 2557, the December 2020 order remanding the five surviving actions, the source that names the Allstate defendants and records that no discovery had commenced. Read on 2026-09-01.

Two sources could not be reached, and a third was reachable only in a form that cannot be string matched. Allstate's own document-release page, on the host media.allstate.com, did not answer: the connection timed out, and the released pages are not served at any other live source that could be found, which is why no slide text appears anywhere above. The Missouri docket behind the contempt fines is likewise not served at any live source reachable here, so the case caption is unestablished and the Missouri figures rest on the Associated Press report alone. Vermont's executed order is served only as an image-only scan with no text layer, so the quotations were taken from the Department's own published order page, which serves the full text as readable HTML.

Every quotation above was re-pulled from the primary source named beside it and matched as an exact string before publication, and each of those sources was read on 2026-09-01. Anything that would not re-verify was cut rather than softened, and the cuts are listed above instead of being quietly dropped. This page is a record of what documents say; it is not legal advice.

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

More guides