Progressive
The best underwriter in this series discloses the least about the algorithms that make it one
Progressive is the most profitable underwriter Finatune has assessed and the quietest about how it gets there. Its competitive edge has been algorithmic pricing since before the industry had a word for it, and the results are not in dispute: a combined ratio of 87.4 for 2025 against a stated target of 96. What is striking is the asymmetry between performance and disclosure. Where other institutions in this series publish adoption metrics for artificial intelligence, Progressive publishes a product model number. And the one algorithm it does describe in detail is the one it licenses from a third party to value totalled cars, which is also the subject of certified class actions in three states and two settlements. The figures come from Progressive's filings with the United States Securities and Exchange Commission. Because Progressive reports on the calendar year, the governing annual report covers 2025 while the operating picture here is the quarter ended 30 September 2026, and both are cited.
87.4
Full-year 2025 combined ratio
Against a stated companywide target of 96
12.6%
2025 underwriting profit margin
On a target of at least 4 percent, worth $10.2 billion pretax
40.5M
Policies in force
Up 6 percent at 30 September 2026
$83.2B
Net premiums written in 2025
An increase of $8.8 billion on the prior year
70,000
Employees at 31 December 2025
21%
Emissions cut against 2022
Against an aspirational 40 percent by 2030
Section 2
The challenge: an algorithmic advantage and an algorithmic liability
Every insurer in this series prices risk. Progressive built its business on pricing it more finely than anyone else, and on being willing to write the drivers its competitors would not once it could tell them apart. Underwriting discipline at Progressive is not a department, it is the entire proposition, and risk management is expressed as a segmentation problem rather than an exposure limit. The 2025 result was a combined ratio of 87.4 and a 12.6 percent underwriting profit margin on a stated target of at least 4 percent, which is to say the company beat its own profitability floor by a factor of three.
The mechanism is data nobody else has in the same volume. Snapshot, offered through hardware and mobile-app versions in every state except California, has been collecting driving behaviour for longer than any comparable programme in the United States, and the company adds commercial variants in Smart Haul and Snapshot ProView. What that produces is a rating structure that can separate drivers who look identical on a credit score and an application form. Third-quarter 2026 renewal applications rose 9 percent while new business applications fell slightly, which is what retention looks like when the price matches the risk.
Two facts sit uncomfortably together. Progressive's advantage is that its algorithms price better than its competitors' algorithms. And its largest consumer litigation concerns an algorithm it licenses from Mitchell International to value totalled vehicles, which plaintiffs say applies an arbitrary downward adjustment. As of 30 September 2026 that claim is certified or conditionally certified in Colorado, North Carolina and Ohio.
The third-quarter numbers add a complication that has nothing to do with litigation. Net premiums written grew 4 percent against 12 percent for the prior full year, which the company attributes to its own rate reductions in what it calls an extremely competitive marketplace. The expense ratio fell 4.5 points in the quarter while the loss ratio rose 3.7, so the margin held and the growth did not. A company that wins by pricing accurately has limited room to win when the market prices below it.
Section 3 Β· Score: 80 / 100
Technology integration
This score is lower than the quality of the technology and higher than the quality of the disclosure, and the gap between those two things is the finding. Progressive runs what is probably the longest-lived machine learning application in American personal lines: telematics-based rating, continuously refined, feeding a segmentation structure it revises as numbered product models. The third-quarter shareholder letter describes the enhanced segmentation of the latest personal auto product model, numbered 9.0, and a new 8.4 core auto model that the company says adds meaningful segmentation capability. That is the whole of the public description. For a company that is arguably the most successful insurtech in the market without using the word, it is remarkably little.
What the filings do establish is the data position. Snapshot operates in 49 states through two delivery mechanisms, and the commercial programmes extend the same approach to fleets. That is a data governance problem as much as a modelling one, since behavioural driving data is among the most sensitive categories an insurer holds and the regulatory treatment of it varies state by state. On the customer-facing side the company holds four patents for the implementation of chatbots in online quoting and servicing, expiring in 2038 or later, which places its natural language processing work roughly a decade ahead of the current industry conversation.
Where Progressive does discuss modern artificial intelligence, it does so defensively. The 10-K risk factors address what the company calls Advanced AI, naming both generative and agentic AI directly, and warn that such systems might produce or reveal datasets that are flawed or insufficient or contain biased information, with the attendant risk of unfairly discriminatory outcomes. The same section acknowledges that competitors may adopt these technologies faster. There is no disclosed AI agent deployment, no named platform, and no large language model programme described in the filings, which is a different posture from every other institution scored above 85 in this series.
The reason this still earns 80 rather than less is that the modelling is load-bearing and verifiable through results rather than announcements. A combined ratio of 87.4 in a year when much of the industry struggled is not a marketing claim. But Finatune scores disclosed evidence, and on that basis Progressive is penalised for a choice it has evidently made deliberately. AI governance is the one area where the company is explicit, noting that the National Association of Insurance Commissioners has adopted AI guiding principles and a model bulletin, and that nearly half of all departments of insurance have adopted that bulletin. A company this exposed to algorithmic scrutiny may simply have concluded that saying less is the safer strategy.
Technology snapshot
- βΈSnapshot telematics in every state except California, via hardware and mobile app
- βΈSmart Haul and Snapshot ProView extending telematics to commercial fleets
- βΈPersonal auto product model 9.0, with a new 8.4 core auto model alongside it
- βΈFour patents on chatbots in online quoting and servicing, expiring 2038 or later
- βΈAdvanced AI named in the risk factors, covering generative and agentic systems
- βΈNo disclosed AI platform, adoption metric or language model programme
Scorecard
Section 4
Digital transformation timeline
2008 -- The emissions record begins
The first year in Progressive's published Scope 1 and 2 emissions series, which the company now reports annually alongside an intensity measure expressed per million policies in force.
2020 -- A carbon neutrality goal is set
Progressive establishes what it describes as an aspirational goal of carbon neutrality by the end of 2025 for Scope 1 and 2 emissions, which it says will help set the path to net zero for those scopes in the following decade.
2022 -- The base year for the 2030 target
The year against which Progressive measures its aspirational goal of reducing total Scope 1 and 2 emissions by 40 percent by 2030.
2024 -- The best year in company history
Management describes 2024 as the most successful year in the history of Progressive, with an underwriting profit margin of 11.2 percent, or $8.0 billion pretax.
2025 -- A harder year, and a better one
Progressive writes $83.2 billion of net premiums, $8.8 billion more than the prior year, ends with 38.6 million policies in force, and posts a combined ratio of 87.4 at a 12.6 percent underwriting margin worth $10.2 billion pretax. Recurring investment income reaches $3.6 billion, up 27 percent.
2025 -- Neutrality, reached with offsets
The target year arrives with a 21 percent reduction in Scope 1 and 2 emissions against 2022 and a further decrease of over 4 percent against 2024. The company retires carbon offset credits equal to 32,250 metric tons of carbon dioxide equivalent, matching its market-based emissions for the year.
2026 -- A class action denied certification
On 15 April an Illinois federal judge declines to certify a class of Progressive customers alleging that a downward pricing adjustment underpaid their total-loss claims, finding their circumstances too varied to resolve together.
2026 -- Growth moderates, margin holds
Third-quarter net premiums written grow 4 percent against 12 percent for the prior full year, with the combined ratio at 88.7 and 87.5 year to date. Policies in force reach 40.5 million, up 6 percent, and quarterly media spend rises 18 percent.
2026 -- The Colorado settlement
A $15.24 million class settlement over the total-loss valuation adjustment receives preliminary approval, with a deadline of 5 November for class members to object or exclude themselves.
Section 5 Β· Score: 92 / 100
Financial performance
Full-year 2025 was exceptional on every measure the company sets for itself. Net premiums written reached $83.2 billion, an increase of $8.8 billion, with policies in force ending the year at 38.6 million. The underwriting profit margin was 12.6 percent against a companywide target of at least 4 percent, producing $10.2 billion of pretax underwriting profit against 11.2 percent and $8.0 billion the year before. Net premiums earned were $81.661 billion on total revenues of $87.671 billion, recurring investment income was $3.583 billion, and the combined ratio finished at 87.4.
The third quarter of 2026 shows a company whose margin is intact and whose growth has slowed. Net premiums earned rose 5 percent to $21.869 billion while net income fell 3 percent to $2.542 billion, or $4.37 per share, with $170 million of pretax net realized losses on securities doing most of the damage. The combined ratio improved to 88.7 from 89.5, and stands at 87.5 year to date. Personal Lines, 89 percent of companywide net premiums written, delivered a 88.0 combined ratio on $19.9 billion of premium. Underlying cash flow remains strong, but the revenue line is decelerating.
$4.37
Third-quarter diluted EPS, down from $4.45
87.5
Combined ratio year to date through September
72.8
Personal property combined ratio
For the quarter, against 76.3 year to date
$37.2B
Total capital at year-end 2025
After $13.50 per share in annual-variable dividends
The capital return tells you what kind of company this is. Progressive paid an annual-variable dividend of $13.50 per share plus a quarterly dividend of $0.10, and repurchased only 0.7 million shares at an average cost of $237.44, ending 2025 with $37.2 billion of total capital. The variable dividend is the mechanism: rather than committing to a buyback programme, the company returns what the year produced and keeps the rest as equity against future growth. For a business whose premium volume can expand 12 percent in a year, that flexibility is worth more than a predictable payout, which is also why the dividend is not a yield story and the market capitalisation does not behave like one.
The reservation that keeps this at 92 rather than higher is concentration of outcome in a single underwriting cycle. Progressive's balance sheet carries a combined premiums-to-surplus ratio of 2.9 to 1 across its insurance companies, with risk-based capital ratios in excess of minimum requirements, so liquidity is not the question. The question is what happens to a 12.6 percent margin when competitors finish repairing their own books and compete on price, which is precisely what the third quarter describes beginning to happen.
Section 6 Β· Score: 82 / 100
Workforce and human capital
Progressive employed about 70,000 people as of 31 December 2025, and reported an annualised retention rate of 90 percent, up one point on the prior year. In 2025 it filled over 81 percent of its open positions above entry level by hiring from within, which for a workforce of this size is an unusually high internal mobility rate and suggests the training function works. The company also runs career development boot camps, two of which are named for information technology programmer and analyst roles.
The employee resource group disclosure is the part Finatune has not seen presented this way elsewhere. Nearly 44 percent of Progressive people belong to at least one such group, and the company reports the retention differential directly: 94 percent for members against 87 percent for non-members. Publishing a seven point gap invites the obvious question of causation, and Progressive publishes it anyway, which is a more useful disclosure than a participation rate on its own would be.
This is a growing workforce in a series full of shrinking ones, and the growth is not incidental to the strategy. A company that competes on the precision of its rating models needs the people who build them, and 81 percent internal promotion above entry level is how an organisation retains that knowledge rather than rehiring it.
What is missing is cost and composition. Progressive discloses neither its total workforce cost nor the split between actuarial, engineering and claims roles, and the proxy's compensation disclosure covers executives rather than the base. The median employee's total annual compensation was $86,906 against the chief executive's $17,705,924, a ratio of 204 to 1. The score of 82 reflects strong, specific and voluntarily granular retention disclosure, held below the mid-eighties because a 10 percent annualised attrition rate is ordinary rather than excellent and because the composition of a workforce this technical is not described at all.
Section 7 Β· Score: 72 / 100
Regulatory compliance
This is the lowest score in the report, and the exposure is consumer litigation rather than prudential enforcement. Progressive holds licences in all 50 states and is supervised by state insurance departments in each of them, with a combined premiums-to-surplus ratio of 2.9 to 1 and risk-based capital ratios in excess of minimum requirements at year-end 2025. There is no federal consent order, no capital restriction and no finding of institutional misconduct. Regulatory compliance for a personal lines insurer is instead tested one claims practice at a time, across fifty jurisdictions, by plaintiffs rather than supervisors.
The pattern in the certified cases is consistent and it is about valuation methodology. As of 30 September 2026, lawsuits were certified or conditionally certified as class or collective actions alleging that Progressive improperly values total loss claims by applying a negotiation adjustment in Colorado, North Carolina and Ohio; improperly calculates basic economic loss for wage loss coverage in New York; improperly reduces or denies personal injury protection benefits when medical expenses are paid initially by health insurance in Arkansas; and improperly fails to process and pay personal injury protection claims on time in Texas. The company notes that other insurance companies face many of these same issues, which is true and does not reduce the exposure. Model risk management is usually discussed as a prudential concept; here it is the subject of the docket.
$61M
Michigan total-loss settlement
Agreed by Progressive Michigan and Progressive Marathon
$15.24M
Colorado total-loss settlement
Preliminarily approved, with a 5 November 2026 opt-out deadline
4 states
With certified class actions at 30 September 2026
Plus the three-state total-loss valuation class
Outcomes have varied, which is itself informative. On 15 April 2026 an Illinois federal judge declined to certify a total-loss valuation class, finding the plaintiffs' circumstances too different to resolve in one case, while Colorado produced a $15.24 million settlement over the same mechanism and Michigan a $61 million settlement over withheld fees. The common thread is a third-party valuation tool licensed from Mitchell International whose downward adjustment to comparable vehicle prices plaintiffs say is arbitrary. A score of 72 reflects clean prudential standing against a recurring, multi-state and partly unresolved challenge to the way the company's software decides what a car was worth.
Section 8 Β· Score: 60 / 100
ESG and sustainability
Progressive reached its 2025 carbon neutrality goal, and the way it reached it is why this score is the second lowest Finatune has awarded. The goal, set in 2020, covered Scope 1 and 2 emissions and was described from the outset as aspirational. The actual reduction achieved by 2025 was 21 percent against a 2022 base year, with total Scope 1 and 2 emissions falling a further 4 percent against 2024. Neutrality was then declared by retiring carbon offset credits equal to 32,250 metric tons of carbon dioxide equivalent, which matches the company's market-based emissions for the year rather than its location-based figure of 64,979 tons.
The forward commitments follow the same pattern. A 40 percent reduction in Scope 1 and 2 by 2030 against 2022 is also labelled aspirational, and at 21 percent the company is slightly over half way with four years left. A separate goal commits to striving for 75 percent renewable energy by 2027 across owned buildings and data centres. There is no Scope 3 target of any kind, and no element of the programme has been validated by the Science Based Targets initiative, which places Progressive behind every institution scored above 70 on this dimension in this series.
32,250
Tons of offsets retired for 2025
Matching market-based rather than location-based emissions
64,979
Location-based Scope 1 and 2 emissions
Against 32,250 tons on a market-based measure
None
Scope 3 target or science-based validation
The programme covers Scope 1 and 2 only
What the programme does have is assurance, and that is the reason this is 60 rather than lower. Progressive engaged PricewaterhouseCoopers to provide assurance over its Scope 1 and Scope 2 emissions for the year ended 31 December 2025, and it publishes the underlying figures including an intensity measure expressed per million policies in force, at 1,683 tons location-based and 835 market-based. It also updated its emissions intensity factors in 2025 to United States Energy Information Administration standards. That is honest, audited, specific reporting of a narrow and offset-dependent commitment, which is better than unaudited reporting of an ambitious one. On climate as a liability rather than a footprint, the filings are candid that climate change may be contributing to the increase in frequency of severe weather events and may be exacerbating their impacts.
Section 9 Β· Score: 66 / 100
SDG alignment
Progressive's development alignment is real where it overlaps the business and thin where it does not. Usage-based pricing has a defensible claim on road safety, because a programme that prices on how a person actually drives rewards the behaviour that prevents crashes, and the company has been running it at scale for longer than anyone. The workforce disclosures support a second claim with published numbers behind them. Beyond those, the contribution is philanthropic rather than structural.
Select a goal to view supporting evidence
The honest limit is that none of this is organised as development work and the company does not present it that way. There is no counted target, no published goal tied to an external framework, and no measurement of outcomes for anyone outside the policyholder base. The score of 66 reflects genuine alignment in the safety and employment dimensions, with no apparatus around it.
Section 10 Β· Score: 89 / 100
Competitive positioning
Progressive's position rests on a cost and accuracy advantage rather than on scale or switching costs, which makes it both impressive and more fragile than it looks. In commercial auto the gap is quantified in the company's own disclosure: Progressive finished the third quarter of 2026 with 9 percent net premiums written growth at a 92.6 combined ratio, against an estimated industry combined ratio of approximately 104 for the first half of 2026 excluding Progressive. A twelve point underwriting gap against the rest of the market is not a marketing position, it is a different business model.
Personal lines shows the same pattern with less headroom. Personal Lines net premiums written of $19.9 billion grew 4 percent at a combined ratio of 88.0, improving 1.5 points, and the personal vehicle business grew policies in force 7 percent at an 88.7 combined ratio described as better than target. Personal property, historically the troubled part of the portfolio, returned a 72.8 combined ratio for the quarter and 76.3 year to date, as the company continued selectively increasing availability of those products rather than retreating from them.
The reservation is the soft market now arriving. Progressive attributes its slower growth to its own rate reductions as competitors take rate and underwriting actions, and auto insurance has almost no switching friction, so a competitor willing to accept a worse combined ratio can buy policies from a company unwilling to. The advantage survives that only if the segmentation genuinely identifies risks others misprice, which is exactly what model 9.0 is for and exactly what the company will not describe. A score of 89 rather than the low nineties reflects a demonstrated advantage in a market structure that does not protect it.
Section 11 Β· Score: 88 / 100
Brand and reputation
Flo has been selling Progressive insurance since 2008 and is one of the most durable advertising assets in American financial services, in a category where brand recall translates directly into direct-channel quote volume. The company is explicit that this is a purchase rather than an accident: quarterly media spend rose 18 percent year on year, and management states it will spend on media to spur growth as long as that spend remains efficient. That is an unusually plain description of brand as a variable cost with a measurable return.
The reputational exposure runs the other way from the advertising. A consumer brand built on friendliness and savings is being litigated, state by state, over whether its claims software underpaid people whose cars were destroyed. The amounts are small against $83.2 billion of premium, but total-loss claims are the moments when a policyholder finds out what the product actually was, and the allegation is specifically that a piece of licensed software quietly reduced what they were owed. Progressive's own framing is that its success in claims has been grounded in its guiding principles for more than twenty years, which is the claim the litigation tests.
+18%
Quarterly media spend growth
Described as continuing while the spend remains efficient
2008
Year the Flo campaign began
Among the longest-running positions in US insurance
+9%
Third-quarter renewal applications
While new business applications fell slightly
204:1
Chief executive to median employee pay ratio
On median employee compensation of $86,906
Section 12 Β· Synthesis
The Finatune lens
Progressive scores 92 on financial performance, which is near the top of this series, and it earns that on a measure the company set for itself and then beat threefold. A 12.6 percent underwriting profit margin against a target of at least 4 percent, $10.2 billion of pretax underwriting profit, $83.2 billion of net premiums written up $8.8 billion, 38.6 million policies in force up 10 percent, and a combined ratio of 87.4. No other institution assessed here outperforms its own stated profitability floor by that distance.
Technology at 80 is the score Finatune expects to be argued with, and the argument is worth having. Progressive's actual capability is probably stronger than several institutions scored higher, because telematics-based rating at this duration and scale is a harder thing to build than a deployed chatbot. But the disclosure is a product model number and a defensive risk factor. Allstate, in the same sector, published its platform architecture and six counted adoption metrics and scored 90. Finatune scores evidence, and a company that chooses not to provide it accepts the consequence.
The shape of this report is a company that is excellent at the thing it will not explain, and exposed on the one piece of software it has had to explain in court. Segmentation model 9.0 produces an 87.4 combined ratio and gets one sentence. The Mitchell valuation adjustment produces $76 million of settlements and hundreds of pages of litigation.
Regulatory at 72 and ESG at 60 are the two weak dimensions and they fail in different ways. The regulatory exposure is real but ordinary for a personal lines insurer: no consent orders, no capital restrictions, clean risk-based capital, and a steady docket of claims-practice class actions that the company correctly notes its competitors also face. The ESG position is weaker than it first appears, because a neutrality claim met by retiring offsets equal to market-based emissions, with no Scope 3 target and no science-based validation, is a narrower achievement than the headline suggests. The PricewaterhouseCoopers assurance is what separates it from the bottom of the series.
Workforce at 82 deserves more attention than it will get. A 90 percent annualised retention rate is unremarkable on its own, but 81 percent of above-entry positions filled internally is not, and publishing a seven point retention differential between employee resource group members and non-members is the kind of disclosure that only a company confident in the number makes. For an organisation whose competitive moat is accumulated modelling knowledge, internal mobility at that rate is a strategic asset rather than a human resources statistic.
What decides the next three years is whether segmentation holds its edge in a soft market. Third-quarter net premiums written grew 4 percent against 12 percent for the prior year, on the company's own rate reductions, while the margin improved. That is the right trade for one quarter and an uncomfortable one for twelve. If model 9.0 genuinely finds risks the market misprices, Progressive keeps writing profitable business while competitors buy unprofitable growth. If it does not, a company with no switching costs, no structural moat and a 4 percent growth rate becomes an ordinarily good insurer with an extraordinary history.
Finatune scorecard
Sources and disclosures
Where these numbers come from
Every financial and operational figure in this report comes from the filings listed below, except where identified otherwise in the text. Because Progressive reports on the calendar year, both annual and quarterly sources are cited, and the period is identified wherever a figure appears. Emissions data and the assurance over it, litigation certification and settlement details drawn from court and industry reporting, and the status of any settlement approval are identified at the point of use.
- Annual Report on Form 10-K
For the fiscal year ended 31 December 2025, filed 2 March 2026, including the 2025 Annual Report to Shareholders as Exhibit 13.
- Quarterly Report on Form 10-Q
For the quarterly period ended 30 September 2026, including the litigation note.
- Letter to Shareholders, Exhibit 99 to the Form 10-Q
The third-quarter 2026 shareholder letter, for the period ended 30 September 2026.
- Proxy Statement on Form DEF 14A
Dated 18 March 2026, for the annual meeting held 8 May 2026, with compensation and pay ratio disclosures covering fiscal year 2025.
Resources
Downloads and Tools
Executive Summary PDF
A two page synthesis of the key findings from this report, formatted for circulation to an investment committee or executive team.
Finatune Assessment Framework Template
Apply the same nine dimension assessment framework to any financial institution, with the scoring rubric and evidence requirements for each dimension.
Related content
Continue across Finatune
Insurance
The insurance category in Finatune's Financial Services module, where Progressive competes against the carriers assessed elsewhere in this series.
Data governance
The data governance platforms we cover, the discipline that decides whether behavioural driving data can be used the way a telematics rating model needs to use it.
Business intelligence
The business intelligence tools in our Data and BI module, against which to read a company that revises its pricing structure as numbered product models.
Compliance and RegTech agents
The compliance and monitoring agents Finatune covers, relevant to an insurer operating under the NAIC AI model bulletin in roughly half the states.
AI financial tools
The AI financial tools category in our Software module, for the applied side of what Progressive builds in-house and does not describe.
Finance glossary
Definitions for the insurance terms used in this report, from underwriting to expense ratio.
This report is an independent assessment prepared by Finatune for informational purposes only. It is not investment advice, and not a recommendation to buy, sell or hold any security. Finatune has no commercial relationship with The Progressive Corporation. Financial figures are drawn from Progressive's filings with the United States Securities and Exchange Commission. Because Progressive reports on the calendar year, annual figures are stated as of the fiscal year ended 31 December 2025 and quarterly figures as of the period ended 30 September 2026, as indicated at each point of use. The litigation described in Section 7 is ongoing: certification rulings, settlements and their approval status may change, the Colorado settlement described had received preliminary and not final approval at the time of writing, and nothing here should be read as a prediction of any outcome or as a statement that any allegation has been proven. Settlement amounts and certification outcomes are drawn from the company's filings and from court and industry reporting, and are identified as such in the text. Emissions figures are drawn from Progressive's own sustainability reporting, over which PricewaterhouseCoopers provided assurance for Scope 1 and Scope 2 for the year ended 31 December 2025. Scores are Finatune's own judgement, applied on a consistent framework to every institution in this series, and reasonable analysts will dispute some of them.