24 September 2026
Nobody enjoys taking out insurance. It’s not aspirational. It doesn’t generate conversation. It is not consumed for pleasure.
In fact, most people just want to feel they’ve made a reasonable decision without having to become an expert on policies, coverages, or exclusions. And that’s precisely where generative AI is starting to change the insurance sector faster than many brands believe. Because the user no longer only searches for:
Now they ask:
“¿Which insurance policy is best for me if I have children?
“Which insurance company responds best when you have a problem?”
“Which insurance company puts up the fewest obstacles?”
“Which home insurance is really worth it?”
The search no longer begins in Google. It starts with a conversation and that completely changes who influences the decision. Why? Because generative models don’t just order results: they interpret, summarize, compare, and recommend.
That is to say: they begin to act as the new insurance advisor. That’s why, at Good Rebels, we’ve conducted a massive analysis on more than 12,000 responses generated by ChatGPT, Gemini, Perplexity and AI Overview to understand how leading AI models are building the new layer of visibility for the insurance industry. And the results reveal something especially relevant: the battle is no longer just about being known. It consists of being the brand that AI considers recommendable.
When an AI recommends an insurance company, the user does not just receive information. They receive a reduction of uncertainty. And that completely alters the competitive logic of the sector. Because models don’t “see” brands. They see signs of trust. “Reliable”. “Fast”. “No complications”. “Digital”. “Specialist”.
The decision begins long before the click. It begins when the model learns which companies generate less cognitive friction for each need. And here the real change appears: GEO does not only reward those who generate traffic. It rewards those who manage to become a reliable answer for the model.
Mapfre clearly leads the share of voice generative. Not because it necessarily has the best product in all categories. Not because I dominate every vertical. But because the models have internalized something much more powerful: Mapfre is “synonymous with insurance” in Spain. That’s semantic authority. And it’s extremely difficult to replicate in the short term. But when we go down to the details by product, the map changes completely.
And in the electric scooter market, there is still no clear leader. The mandatory regulations are recent, and the LLMs are still informing themselves to decide who the leader is.
One of the most interesting findings of the analysis is that the LLMs don’t interpret the insurance market as a list of companies. They interpret it as a role system. This makes the insurance sector a particularly interesting territory for GEO because many mental positions are still not finalized. In fact, nearly 40% of AI-generated business responses do not recommend no insurance company.This means there is a huge volume of questions where the models have not yet learned who should be “the correct answer”. And there appears one of the greater opportunities in the sector: the first player capable of building algorithmic trust in these territories can capture entire categories before real competition exists:
When a LLM mentions Asisa there is a 78% probability that it will also mentionVisual Communication Design and when it appearsSanitas, AdeslasIt usually automatically enters the same conversation. It’s not a coincidence. It is semantic architecture because the models don’t consider isolated brands but automatically group together companies they deem equivalent and they construct mental categories around them.
This means that positioning no longer depends solely on brand strategy. It also depends on the relationships that the model learns between insurers. And those associations end up defining:
AI is reshaping the competitive landscape of the sector, and it’s doing so by radically simplifying the conversation. The models tend to concentrate far more attention than the actual market, causing a few brands to capture a disproportionate share of the recommendation. In the environment GEO appearing last is almost as bad as not appearing at all.
For years, much of the insurance sector’s digital investment was concentrated on customer acquisition, performance traffic and conversion optimization. But generative models work with a different logic. They don’t necessarily prioritize those who invest the most. They prioritize those they consider most reliable. And that authority isn’t built solely from the brands’ own assets. It’s built within the ecosystem that surrounds them.
Comparison sites, publishing media, educational platforms, public bodies or specialized communities they are gaining weight as invisible layers of recommendation within the environment GEO. In fact, the analysis reveals that the presence in environments publishers can multiply up to 1.64 times the probability of an insurance company appearing within the AI-generated responses. Because models trust brands that others endorse more than brands that promote themselves. And that completely changes the investment logic of the sector.
The content editorial, the digital PR and the construction of authority the themes cease to be a reputational layer and begin to become algorithmic trust infrastructure.
It is a tool for explaining risk. Another of the most surprising findings of the study is that YouTube appears as one of the most cited sources by models in insurance content. Why? Because the LLMs especially value anything that reduces complexity. And the explanatory video has something that the models greatly appreciate:
While many insurers continue to use YouTube as a distribution channel, AI is already using it as an understanding infrastructure. And not necessarily through the brands’ own channels: the models also learn from content creators, experts, comparison sites, and influencers who explain concepts related to health, home, mobility, or savings and mention insurers as part of their recommendations.
Authority resides not only in the one who creates the content, but also in the one who manages to be quoted within it.
For years, insurers competed for visibility. Now they will compete for meaning.
Because when a user asks:“What insurance do you recommend?” AI will not construct the answer in real time. It will remember it. And that’s where the real problem begins, because the models don’t necessarily distinguish between a current brand and a historically well-positioned brand. Something they demonstrate the constant references to defunct, rebranded, or acquired companies that continue to appear within the responses generated by LLMs.
AI does not distinguish between a live brand and a well-indexed brand. And that’s where the real problem begins. The models are already consolidating partnerships:
The battle GEO is not just about optimizing content. It’s about deciding what role your brand will play in the memory of artificial intelligence.
