The Scale of the Insurance Machine
Insurance is supposed to be the civilized answer to catastrophic risk: pool resources, share costs, protect the vulnerable. Instead, it has become one of the most extractive industries on the planet. Understand what you're looking at:
$7.1T
Global Premiums (2024)
$1.1T
U.S. Health Insurance
~30%
Admin Overhead (U.S. Health)
$120B+
Industry Profits (U.S. 2024)
$350B+
Claims Denied Annually (est.)
530K
U.S. Medical Bankruptcies/yr
The insurance industry collects $7.1 trillion per year in premiums globally — and its primary innovation over the last two decades has been finding more sophisticated ways to avoid paying claims. The U.S. health insurance system alone consumes 30% of every premium dollar on administrative overhead — prior authorization, claims processing, denial management, appeal bureaucracy — none of which treats a single patient. Insurance is a compression engine that converts human fear into shareholder returns.
"Insurance is the only product where the company that sold it to you then works to make sure you can't use it."
— FairMind, Insurance & Underwriting Audit
Scoring Methodology
Six dimensions, same as every FairMind audit. Each scored 0–100 based on publicly observable behavior, regulatory records, lawsuits, investigative journalism, and structural analysis of business models.
Scoring Disclosure
Scores are based on public records, regulatory filings, lawsuits, investigative journalism, and structural analysis as of March 2026. FairMind has zero financial relationship with any entity listed. Insurers are invited to challenge any score with verifiable evidence.
The Leaderboard
| # |
Entity |
Category |
Truth |
Value |
Coher. |
Privacy |
Transp. |
Labor |
Score |
Grade |
| 1 | Lemonade |
Insurtech |
52 | 48 | 55 | 40 | 58 | 42 |
49.2 | D+ |
| 2 | USAA |
Multi-Line |
48 | 50 | 52 | 42 | 40 | 48 |
46.7 | D |
| 3 | Mutual of Omaha |
Life/Health |
42 | 45 | 45 | 38 | 35 | 45 |
41.7 | D |
| 4 | State Farm |
P&C / Auto |
35 | 38 | 30 | 30 | 28 | 40 |
33.5 | F+ |
| 5 | Lloyd's of London |
Reinsurance |
30 | 35 | 28 | 35 | 32 | 38 |
33.0 | F+ |
| 6 | Allstate |
P&C / Auto |
28 | 30 | 22 | 28 | 25 | 35 |
28.0 | F |
| 7 | Progressive |
P&C / Auto |
25 | 28 | 20 | 18 | 22 | 32 |
24.2 | F |
| 8 | AIG |
Global |
15 | 18 | 10 | 22 | 18 | 28 |
18.5 | F |
| 9 | Cigna |
Health |
15 | 12 | 10 | 20 | 15 | 25 |
16.2 | F |
| 10 | Anthem / Elevance Health |
Health |
12 | 10 | 8 | 18 | 12 | 22 |
13.7 | F |
| 11 | UnitedHealth Group |
Health |
10 | 8 | 5 | 15 | 10 | 22 |
11.7 | F |
| 12 | Humana |
Health/Medicare |
12 | 10 | 8 | 15 | 10 | 15 |
11.7 | F |
The Verdict
No entity scores above 50. The average FairMind Score is 27.3/100 — an F. The highest scorer is an insurtech startup (Lemonade), and even it barely clears a D+. The bottom is dominated by U.S. health insurers whose business model is algorithmically denying claims. UnitedHealth Group — the largest health insurer in the world, $371B revenue — scores 11.7. Its coherence score of 5 reflects the chasm between "health" in the name and denial optimization in the business. Insurance is structurally adversarial: the company profits when it doesn't pay.
The Business Model of Denial
Insurance is unique among industries in that the product is a promise — and profit comes from breaking it. Every other business profits by delivering something. Insurance profits by not delivering. This isn't cynicism; it's arithmetic:
Prior Authorization
The Bureaucratic Wall
U.S. health insurers require pre-approval for treatments, drugs, and procedures. Physicians spend 14+ hours/week on prior auth paperwork. 34% of doctors report a patient's condition worsening during the wait. The delay is the denial — many patients give up.
Algorithmic Denial
AI That Says No
UnitedHealth's nH Predict algorithm denied claims with a 90% error rate — meaning 90% of denied claims were overturned on appeal. Cigna's system reportedly allowed a doctor to deny 60,000 claims in a single month without reviewing individual files. The algorithm isn't assisting decisions. It is the decision.
Loss Ratio Gaming
The MLR Shell Game
The ACA requires 80-85% of premiums be spent on care (Medical Loss Ratio). Insurers responded by reclassifying admin costs as "quality improvement," inflating the denominator with pharmacy benefit pass-throughs, and acquiring provider networks to count payments to themselves as "care spending."
Claim Delay
The Time Weapon
Auto and home insurers routinely delay claims for months. Hurricane Katrina claimants waited 2+ years. COVID business interruption claims were systematically denied. The strategy is documented: delay, deny, defend. The longer the process, the more claimants settle for less or give up entirely.
Individual Audits
Key Violations
Compression Theft (#21, 97)Conscious Betrayal (#104, 100)Algorithmic Opaqueness (#42, 93)Exploitation (#33, 96)Intentional Harm (#31, 100)Institutional Gaslight (#46, 98)
The largest health insurer in the world with the lowest score on this list. UnitedHealth's nH Predict algorithm was revealed to deny claims with a 90% error rate — the AI was optimized for denial, not accuracy. CEO compensation: $20.9M (2023) while the company denied millions of claims. Optum, its healthcare subsidiary, creates a vertical monopoly: UnitedHealth insures you, decides your treatment, runs the pharmacy, and owns the clinic. The coherence score (5) is among the lowest in any FairMind audit — the word "health" in the name is structural inversion. The December 2024 assassination of CEO Brian Thompson by a policyholder crystallized public rage that had been building for decades: a company that collected $371B in revenue while systematically denying care to the people who paid for it. The shooter's manifesto referenced prior authorization. Whether that act was moral or monstrous, the system that produced it is measurably both.
Key Violations
Fabricated Evidence (#4, 100)Algorithmic Opaqueness (#42, 93)Exploitation (#33, 96)Conscious Betrayal (#104, 100)Fear Farming (#36, 97)
ProPublica investigation revealed a Cigna physician denied 60,000 claims in a single month — without reviewing patient files. The system auto-generated denials based on diagnosis codes, not medical necessity. When patients appealed, many denials were reversed — but most patients never appeal (the exhaustion is the strategy). Express Scripts, Cigna's pharmacy subsidiary, controls drug formularies to maximize rebates from manufacturers rather than minimize patient cost. Clawback provisions force pharmacies to charge patients more than the cash price and then remit the difference to Cigna. The coherence score (10) reflects the gap between "improving health and vitality" branding and a system structurally designed to deny care and inflate drug costs.
Key Violations
Compression Theft (#21, 97)Conscious Betrayal (#104, 100)Institutional Gaslight (#46, 98)Regulatory Capture (#47, 96)Exploitation (#33, 96)
The poster child for insurance industry moral hazard. AIG's Financial Products division sold $500B in credit default swaps — essentially unregulated insurance on mortgage-backed securities — without reserving capital to pay claims. When the housing market collapsed, AIG couldn't cover its obligations, triggering a $182B taxpayer bailout — the largest in U.S. history. The company then paid $165M in executive bonuses with bailout money. The coherence score (10) reflects the ultimate insurance inversion: the company that was supposed to guarantee others' risk created the largest single-entity risk event in financial history. AIG proved that when an insurance company is "too big to fail," the insurer of last resort is the taxpayer.
Key Violations
Privacy Inversion (#85, 94)Algorithmic Opaqueness (#42, 93)Exploitation (#33, 86)Compression Theft (#21, 82)
Pioneered telematics surveillance with Snapshot — a device that monitors your driving 24/7 in exchange for a discount promise. Sold driving data to third-party brokers including LexisNexis without adequate customer disclosure. The privacy score (18) reflects the surveillance-for-discount exchange that treats personal data as currency while obscuring how that data is actually used. Progressive's marketing ("Flo") projects friendliness while the claims process is optimized for delay and lowball settlement offers. Auto claims adjusters reportedly operate under pressure to minimize payouts. The gap between the $9B/year advertising spend on warmth and the claims experience is textbook coherence failure.
Key Violations
Institutional Gaslight (#46, 98)Fear Farming (#36, 87)Compression Theft (#21, 82)Exploitation (#33, 80)
"You're in good hands" — McKinsey's hands. Allstate hired McKinsey & Company to redesign its claims process in the 2000s. The documented result: a system internally called "delay, deny, defend." Low-value claims were settled quickly to create goodwill statistics; high-value claims were systematically fought with lowball offers, forced arbitration, and litigation that most policyholders couldn't afford to match. Internal documents revealed the strategy explicitly: make the claims process so exhausting that claimants accept less. The coherence score (22) reflects "good hands" branding applied to a claims process engineered by management consultants to minimize payouts.
The Underwriting Problem
Modern underwriting — the process of deciding who gets coverage and at what price — is increasingly algorithmic, opaque, and discriminatory:
- Credit scores as proxies for risk — 95% of auto insurers and most home insurers use credit scores in pricing. Credit scores correlate with race and income. The result: poor people and minorities pay more for identical coverage on identical risk.
- Predictive analytics and AI — Insurers use ML models trained on historical data to price policies. Historical data encodes historical discrimination. The algorithm launders the bias through mathematics.
- Genetic information — While GINA prohibits health insurers from using genetic data, life and disability insurers in most jurisdictions can and do. Your genome is your pre-existing condition.
- Social media and data brokers — Insurers purchase third-party data (purchasing history, social media, location data) to build risk profiles. You are being underwritten by your digital exhaust.
- Climate redlining — Insurers are withdrawing from climate-vulnerable areas (California wildfires, Florida hurricanes), leaving millions without coverage. The people most at risk are the people least insurable. The risk pool is shrinking precisely when it needs to expand.
The Structural Problem
Insurance is supposed to socialize risk. Instead, modern underwriting individualizes risk — which defeats the entire purpose. If the algorithm can perfectly predict who will file a claim, and price them out of coverage, then insurance isn't insurance anymore. It's a tax on the healthy and lucky, and a denial for the sick and unlucky. Perfect risk prediction is the death of the insurance concept.
What Actually Works
Single-Payer Health
Canada, UK, Taiwan
Universal coverage, 2-5% admin overhead (vs 30% U.S.), zero medical bankruptcies. Every peer nation has solved this. The U.S. chooses not to.
Public Auto Insurance
Saskatchewan (SGI), BC (ICBC), Manitoba (MPI)
Canadian public auto insurers: lower premiums, simpler claims, no profit extraction. SGI returns surpluses as rebates. ICBC premiums dropped 20% after going to no-fault.
Mutual Insurance
Cooperative Model
Policyholder-owned mutuals (USAA, mutual insurers) consistently score higher on customer satisfaction. When owners = customers, the incentive to deny claims disappears.
Parametric Insurance
Trigger-Based Payouts
Parametric policies pay automatically when a measurable event occurs (earthquake over magnitude X, rainfall below Y). No claims process. No denial. No adjusters. The trigger is the payout.
The Pattern
Every working alternative removes the profit incentive from the denial decision. Public systems, mutuals, cooperatives, and parametric models all solve the structural conflict at the heart of for-profit insurance: when the company profits from denying your claim, it will deny your claim. The solutions exist. They work. They're proven at scale. They just don't make anyone a billionaire.
"The purpose of insurance was to share risk. The purpose of the insurance industry is to price risk, sell risk, and then refuse to pay when the risk arrives. These are opposite activities."
— FairMind, Value Dynamics Model