Competitors Lost a Third of Their Traffic to AI Overviews. This Travel Site Grew ChatGPT Revenue 270% Instead.
In October 2025, a travel insurance comparison site published a survey of nearly 2,500 of its own customers. Eighty-one percent said they found AI helpful when looking for travel insurance.
The company used the survey to issue a warning. Travelers were asking assistants about visa rules and flight timings, some were getting confident wrong answers, and no policy covers a missed flight caused by bad advice from a chatbot. Double-check with official sources, the company said. Do not let an AI pick your policy.
Then it did the thing a lot of businesses talk about and few actually do. It accepted that its customers were going to ask AI anyway, and made sure that when they did, the answer included its name.
The two numbers
According to the agency that published the case study, the company grew ChatGPT-driven revenue by 270% over six months.
The number sitting next to it is the one that should worry everyone else. Over the same period, competitors that did not adapt lost 34.5% of their traffic to Google's AI Overviews.
Read those together. This is not only a story about upside. It is a story about a market where the same shift that rewarded one company was quietly taking a third of the traffic from the ones standing still. When an AI answer sits at the top of the page, the businesses named inside it gain and the businesses below it lose, and the losing happens whether or not anyone is watching.
What "re-engineered as conversational answers" means
The strategy, as described, was to rebuild content as conversational answers rather than as pages.
A travel insurance marketplace has thousands of policies, each with exclusions, each suited to different trips. A traditional page lists them. A conversational answer says: for a traveler with a pre-existing condition taking a cruise, here is what matters and here is why. That second form is what a model can lift into a reply when someone asks exactly that question.
The company had twenty years of policy knowledge and more than four million customers' worth of decisions. The work was reshaping that knowledge into the form an assistant could quote, and adding the structured data that helps a model identify the primary source for a claim.
Why a comparison site is the hard case
This is worth dwelling on, because a marketplace is one of the businesses most exposed to AI answers. Its entire value is comparison, and comparison is precisely what an assistant now does in the chat window. If the model compares the policies itself and names the insurers directly, the marketplace disappears from the journey.
That the company grew instead of shrinking suggests the model still needs a trusted intermediary for complex, exclusion-heavy decisions, and will cite one if it can find one that answers clearly. The company positioned itself as that source.
The caveats
The figures come from the agency's own case study and have not been audited independently. "ChatGPT-driven revenue" is attributed revenue, which for AI referrals tends to undercount, because many people who receive a recommendation navigate directly rather than clicking. The true effect could be larger, or, if attribution was generous, smaller. The direction is consistent with what the company's own survey said its customers were doing.
The part that transfers
Every business has a version of that October survey. Your customers are already asking assistants about your category. The only open question is whether the answer names you or a competitor.
The travel site's sequence is the useful bit: find out what your customers are doing, in their own words. Accept it. Then reshape what you already know into the form an assistant can quote.
And keep an eye on the second number. A third of your traffic is a lot to lose without noticing.
References
- Squaremouth press room, "More Travelers Are Using AI Without Realizing the Risks," 21 October 2025. squaremouth.com
- Previsible case studies page, documenting the Squaremouth results. previsible.com
- The same Squaremouth release as distributed via PR Newswire, republished by Morningstar. morningstar.com