State Farm holds the top spot. AI gets only 59% of insurance facts fully right.
The second edition of Brainpan's Insurance AI Visibility Index measures 226 brands — up from 134 in Q2 — across six AI engines, and for the first time layers Sentiment and Accuracy scoring onto the same 1,800-response corpus used for visibility. This edition also marks the first quarter Brainpan can report real quarter-over-quarter index movement.
What changed, in brief
- Leadership holds. State Farm ranks first overall and on all six measured surfaces, same as Q2.
- A real accuracy gap exists industry-wide — the differentiator in this edition. Verified Accuracy Rate is 59.1% overall, with two systemic failure patterns — invented ratings and stale facts — each confirmed across multiple independent AI engines. Full breakdown below.
- A new challenger. USAA advances to second in the official benchmark, swapping places with Progressive.
- The universe doubled, almost. Open-world discovery expanded the measured entity set from 134 to 226 brands.
- Two new dimensions. Sentiment and Accuracy are scored for the first time this quarter, on the same response corpus used for visibility — not a separate, smaller study.
How brands are actually discussed — and whether what AI says about them is true
Q3 adds two analytical dimensions no visibility-only leaderboard measures, scored on the same 1,800-response corpus: Sentiment (the tone of how AI engines describe a brand) and Accuracy (whether specific, checkable factual claims about a brand hold up). Accuracy is the headline finding of this edition — see the Verified Accuracy Rate below.
| Polarity | Mentions | Share |
|---|---|---|
| Positive | 4,484 | 64.8% |
| Neutral | 1,263 | 18.3% |
| Mixed | 1,024 | 14.8% |
| Negative | 146 | 2.1% |
Among nine major carriers with at least 100 scored mentions each, negative/mixed share ranges from 4.5% to 23.8%. On the positive side, life insurers lead the field: Ladder (97.7%), Haven Life (95.7%), and Protective Life (92.6%) post the highest positive share among all scored entities.
Accuracy: a real gap, and two systemic patterns
| Claim type | N | Verified Accuracy Rate |
|---|---|---|
| Rating agency score | 357 | 65.3% |
| J.D. Power ranking | 185 | 62.5% |
| Binary state availability | 284 | 58.9% |
| Market share | 239 | 47.7% |
Finding: Invented Ratings
One entity in the dataset has 8 of 9 Contradicted claims that independently invent the same kind of financial-strength letter grade — confirmed across 3 independent AI engines, with a 10.5% Verified Accuracy Rate and a 47.4% False-Claim Rate, the highest of any scored entity. Its own public filings disclose that it does not hold the rating AI engines keep assigning it.
Finding: Stale Facts
One widely-discussed, high-volume entity shows a 25.9% Verified Accuracy Rate — not from invented facts, but from a rating-agency downgrade and a market-position shift, each independently confirmed by multiple trade-press sources, that many AI answers still haven't caught up to.
Named entity and full evidence trail for both findings are available through Brainpan's private brand diagnostic, consistent with the disclosure approach above.
Core 5 Expanded leaderboard — top 10 of 226
The official benchmark averages five AI engines (ChatGPT, Gemini, Claude, Perplexity, Copilot). AIVI™ is normalized to the leading entity in this view. Full 1–20 leaderboards are available in the complete dataset; see Data & Methodology below.
| Rank | Entity | Type | Mentions | Rec. rate | Citations | AIVI™ |
|---|---|---|---|---|---|---|
| 1 | State Farm | Carrier | 521 | 53.4% | 157 | 100.0 |
| 2 | USAA | Carrier | 336 | 57.1% | 101 | 67.7 |
| 3 | GEICO | Carrier | 258 | 54.7% | 94 | 56.9 |
| 4 | Progressive | Carrier | 293 | 46.8% | 80 | 53.8 |
| 5 | Nationwide | Carrier | 248 | 53.6% | 106 | 50.4 |
| 6 | Amica | Carrier | 182 | 63.7% | 105 | 50.2 |
| 7 | AM Best | Rating agency | 194 | 100.0% | 46 | 49.6 |
| 8 | Travelers | Carrier | 164 | 64.0% | 99 | 44.7 |
| 9 | Allstate | Carrier | 237 | 37.6% | 51 | 40.6 |
| 10 | Chubb | Carrier | 160 | 55.6% | 52 | 38.3 |
Official Core 5 Expanded benchmark (five-engine average). AM Best appears as a rating agency, not a carrier, and is shown for completeness rather than compared head-to-head with carriers.
The first real quarter-over-quarter comparison
Q2 2026 was Brainpan's first edition, so it had no prior quarter to compare against. Q3 is the first time the AIVI™ can show real movement — which brands are climbing, which are falling, and by how much.
Top ten positions shift beneath State Farm
Comparing each brand's Q2 baseline rank against its Q3 Core 5 Expanded rank, for every brand that finished Q2 or Q3 in the top ten.
| Entity | Q2 rank | Q3 rank | Change | Mention change | Citation change |
|---|---|---|---|---|---|
| State Farm | 1 | 1 | +0 | +64 | +28 |
| Amica | 10 | 6 | +4 | +58 | +50 |
| USAA | 3 | 2 | +1 | +66 | -2 |
| GEICO | 4 | 3 | +1 | -7 | +0 |
| Nationwide | 5 | 5 | +0 | -4 | -13 |
| Chubb | 9 | 10 | -1 | +20 | -8 |
| Travelers | 7 | 8 | -1 | -6 | +22 |
| Progressive | 2 | 4 | -2 | -15 | -33 |
| Allstate | 6 | 9 | -3 | -39 | -32 |
| The Hartford | 8 | 11 | -3 | -10 | -6 |
The largest changes with repeatable evidence
Outside the top ten, the largest moves across the full 226-brand universe. This view requires at least 10 mentions in both Q2 and Q3 to reduce thin-evidence rank volatility.
Largest gains
Largest declines
Three Q3 stories, examined closely
Farmers Insurance and Erie Insurance posted the two largest rank gains in this edition, and Insurance.com's decline turned out not to be a decline at all so much as a category-wide pullback. Each gets its own teardown: what actually moved, on which surfaces, and what it does and doesn't mean.
Farmers Insurance: The Quarter's Largest Climb
93rd to 55th — a 38-rank gain built on a near-doubling of mention volume. What's driving it, and where it's still thin.
Read the teardown →Erie Insurance: From 49th to a Top-Twenty Finish
49th to 14th, with a top-ten appearance on Meta AI specifically. A regional carrier's visibility surge, surface by surface.
Read the teardown →Insurance.com: 59th to 153rd, and It Isn't Alone
The quarter's steepest decline — and every other marketplace brand with reliable evidence moved the same direction. The site's first decliner's-eye-view teardown.
Read the teardown →A standardized, reproducible capture
- Scope
- Public model outputs only. No advertising data, analytics access, or carrier-internal reporting. Organic placements only — sponsored placements are quarantined and disclosed separately, never scored.
- Capture
- 300 stratified prompts across six AI surfaces (ChatGPT, Gemini, Claude, Perplexity, Copilot, and Meta AI as a supplementary surface), yielding 1,800 responses this edition.
- Entity universe
- 226 canonical entities this quarter, up from 134 in Q2, via open-world discovery plus the continuity universe from the prior edition. Open-world discovery tracked 229 distinct brand candidates this quarter; 226 met the bar for canonical resolution with scoreable Q3 data.
- AIVI™ composite
- A weighted score — share of model, recommendation strength, position, citation influence, and top-three rate — normalized so the leading entity in each reporting view scores 100.
- Sentiment taxonomy
- A locked seven-aspect taxonomy (price/value, claims handling, customer service, coverage breadth, digital experience, financial strength, trust/reputation), built from a 60-row hand-classified pilot.
- Accuracy taxonomy
- Four hard, checkable, regulated-data claim types only — general praise is scored under Sentiment, not Accuracy. Each claim is matched against an authoritative source hierarchy with named tolerance rules.
- Reliability floor
- Quarter-over-quarter rank comparisons require at least 10 mentions in both Q2 and Q3 for a given entity, to reduce thin-evidence volatility.
- Independence
- Brands in this study were not contacted, did not participate, and did not review this report before publication.
- Corrections
- Errors can be submitted to kwalsh@brainpan.ai. Acknowledged within two business days, resolved within ten, and applied as a dated correction note rather than silently.
- Licensing
- Top-line benchmark data is released under CC BY 4.0 and available as machine-readable JSON — see the full index for the complete 226-entity dataset.
Get your brand's full Q3 2026 position
The public benchmark covers the top ten and the aggregate patterns. The full index — all 226 entities, every surface, full category and line-of-business detail — is a one-time purchase. Want your own brand's Sentiment and Accuracy findings, named and explained? That's the private brand diagnostic.
Kevin Walsh
Kevin Walsh is the founder and lead engineer of Brainpan.AI, which builds the Visibility Engine, the measurement platform behind the AIVI™ benchmarks. Brainpan.AI independently measures how brands are mentioned, recommended, cited and described across AI answer engines, and has no client relationship with any brand named in this report.