We just published 5W's Auto Insurance AI Visibility Index — 288 real answers from ChatGPT, Claude, Gemini, and Perplexity to questions like "best car insurance company" and "cheapest car insurance." State Farm came out on top. But what stuck with me wasn't the leaderboard. It was what the study says about where trust actually gets built now.
An industry I've watched build brands the old way
I've spent two decades in this business watching categories build recognition through sheer advertising force. Auto insurance is maybe the best example there is — a gecko, a woman named Flo, a chaos agent named Mayhem. These aren't just ads, they're culture. Everyone can name them whether or not they've ever shopped for a policy.
So when our research team brought me the numbers on this study, I wasn't surprised that State Farm led at 65.2%, with Progressive and Nationwide right behind in a statistical near-tie at 61.4% each. What surprised me was the layer underneath.
Why don't AI engines trust the same voices we spent decades building?
When ChatGPT, Claude, Gemini, and Perplexity cited a source behind their answers, it wasn't the insurer's own website nine times out of ten. It was NerdWallet, cited in 53.3% of answers with evidence. Or insurance.com, at 52.7%. Or Forbes Advisor, at 51.6%. Compare that to the best-performing carrier's own site — Nationwide, at just 13.6%. Several major carriers were cited from their own domains in exactly zero of our measured answers.
That's the part I keep coming back to. Decades of broadcast advertising built enormous brand recognition. It did almost nothing to build the kind of citable, structured, third-party-referenced content that gets a company's own site treated as a trustworthy source by an AI system. Those are two different jobs, and most of this industry has only ever done one of them well.
It also isn't just a media-spend problem. It's a content-architecture problem. The publishers winning these citations write comparison content built specifically to answer the exact questions consumers ask — structured, specific, and updated constantly. Most insurer websites are built to sell a policy once someone has already decided on the brand, not to answer the open-ended research question that comes first. That's the gap AI search has exposed.
Is this pattern unique to insurance? I don't think so.
Our AI Visibility Index series has now measured this same gap in pet insurance, credit cards, home security, and dozens of other categories across more than 60 industries: the brand spending the most doesn't automatically become the brand AI trusts as a source. I think this is the single most important thing communicators need to understand about the next decade of our industry. The channel isn't changing. The definition of authority inside that channel is.
One more thing worth noting: the four AI systems didn't agree with each other. Perplexity named State Farm in 95.7% of its answers. Gemini named the same brand in just 39.1%. If you're only watching one chatbot, you're not measuring your AI visibility — you're measuring a quarter of it, at best.
You can read the full study, including the complete carrier leaderboard and engine-by-engine breakdowns, on 5W's research page: the Auto Insurance AI Visibility Index. You can also browse the rest of our findings in the AI Visibility Index series, which now covers more than 60 categories.