AI Search & GEO Insurance Statistics 2026: AI Overviews, Citations and Consumer Adoption

Insurance has always been a search-first category. People type a worry into Google, land on a comparison site, and buy from whoever the panel puts in front of them.

That chain is being quietly rerouted. A growing share of insurance questions now get answered before anyone reaches a website, and a separate share get answered inside ChatGPT, where there is no panel and no quote form.

The data is uneven. There is decent UK evidence on how many people use AI for money tasks, and almost nothing published on how much traffic AI search actually sends to insurance sites. Where the public record is thin, this piece uses original SERP and citation analysis run in Ahrefs, clearly labelled, so you can see the method and disagree with it.

Key AI search and GEO insurance statistics for 2026

• 45% of UK adults who use AI for money tasks use it for insurance comparison, making it the fourth most common financial use of AI, according to Lloyds Banking Group [2]
• 28 million UK adults, 56% of the adult population, used AI to manage money in the previous 12 months (Lloyds Banking Group, November 2025) [2]
• 17% of UK consumers used AI when shopping for or switching a financial product in the past year (FCA, n=5,026) [3]
• 93% of the 60 highest-volume question-phrased UK insurance queries return a Google AI Overview, against 14% of the 100 highest-volume UK insurance keywords overall (author analysis, Ahrefs UK SERP data, August 2026) [1]
• AI Overview presence ranges from 73% on business insurance keywords to 15% on motor across a 360-keyword UK sample (author analysis, Ahrefs, August 2026) [1]
• Which? has been cited 193,684 times across five AI surfaces, against 6,867 for Aviva and 841 for Direct Line (author analysis, Ahrefs, August 2026) [1]
• 26% of UK consumers trust general-purpose AI tools such as ChatGPT for financial advice (FCA Mills Review, July 2026) [4]
• 95% of UK insurance firms were already using AI in 2024, the highest of any financial sector (Bank of England and FCA, 118 firms) [5]
• Only 40% of property and casualty insurance leaders say AI is meeting expectations, and 42% have not measured its results at all (Capgemini, 2026) [6]
• Willingness to use AI falls from 36% to 20% as it moves from recommending to acting autonomously (FCA, April 2026) [3]

How often do UK insurance searches trigger an AI Overview?

It depends almost entirely on how the query is phrased. Question-form insurance searches almost always get an AI Overview. Brand and quote-intent searches almost never do.

I pulled the 100 highest-volume UK keywords containing "insurance" from Ahrefs in August 2026 and checked which SERPs carried an AI Overview. Fourteen did. I then pulled the 60 highest-volume UK insurance keywords phrased as questions, and 56 carried one, or 93% [1]. Stripping out National Insurance queries, a tax topic rather than a product one, the rate barely moves: 38 of 41 remaining questions, also 93%.

That gap is the single most useful thing an insurance marketer can know right now. The money keywords, "car insurance", "compare the market car insurance", "cheap car insurance", still return a clean commercial SERP. The advice keywords, "what does third party insurance cover", "why has car insurance gone up", "how much is pet insurance", are already answered on the page.

The implication is uncomfortable rather than catastrophic. Insurers and brokers spent a decade building advice hubs as cheap top-of-funnel entry points into quote journeys, and it is that entry point being absorbed, whilst the expensive bottom-of-funnel keywords are for now untouched.

Which insurance lines are most exposed to AI Overviews?

Business insurance is by far the most exposed. Motor is the least. The gap between them is roughly five to one.

For each of six lines I took the 60 highest-volume UK keywords containing that line's term and recorded AI Overview presence, giving a 360-keyword sample. Method and date are stated so the analysis can be repeated.

Source: author analysis of Ahrefs UK keyword data, August 2026 [1]

Two things drive the spread. Motor and travel have enormous branded and navigational head terms that crowd out the informational tail, whilst business insurance has almost no consumer brand gravity. And life and business insurance generate genuine "what is" and "do I need" questions, which is precisely the query shape Google answers itself.

The risk is therefore not evenly distributed inside a single business. A composite insurer's commercial lines content is materially more exposed than its motor book, and a specialist SME broker running purely on definitional content is more exposed than either.

Do not read 73% as "73% of business insurance traffic is gone". Presence is not click loss, and the two should never be conflated.

Which domains do AI assistants cite for insurance questions?

Consumer champions, not insurers. Which? and MoneySavingExpert are cited an order of magnitude more often than any insurance brand.

Ahrefs tracks how often a domain appears as a citation across five AI surfaces. These counts are sitewide rather than insurance-only, so treat them as a measure of a domain's standing as an AI-citable source in the UK rather than a clean insurance share.

Source: author analysis, Ahrefs Site Explorer, United Kingdom, August 2026 [1]

Which? attracts 230 times as many AI citations as Direct Line, and MoneySavingExpert 18 times as many as Aviva. Editorial authority, not brand size or advertising spend, is what these systems reward.

The platform splits matter too. Which? is heavily ChatGPT-weighted, with 61% of its citations on one surface, whilst MoneySupermarket is spread almost evenly across all five. Compare the Market registered just two Gemini citations against 7,802 in AI Mode, a reminder that per-platform numbers at this granularity are fragile.

The regulator outperforming most insurers is the detail I would put in front of a board. The FCA's 7,844 citations beat Aviva and dwarf Direct Line, because AI systems reach for institutional, non-commercial explanation when a question is about rules, rights or definitions. Content that reads like a sales page loses those queries to a regulator or a consumer group every time.

How many people actually use AI to compare insurance?

More than most insurers assume, but fewer than the AI hype suggests. UK estimates cluster between roughly one in six and one in five adults for financial tasks, with insurance a consistently popular use case.

Lloyds Banking Group's Consumer Digital Index, published in November 2025, found 28 million UK adults, 56% of the adult population, had used AI to manage money in the previous 12 months, with 45% of those users applying it to insurance comparison [2]. ChatGPT was the platform for six in ten, and one in three used it weekly or more.

The FCA's own research, run with Yonder among 5,026 consumers in April 2026, is more conservative. It found 16% currently use AI for at least one personal finance activity, and 17% used AI when shopping for, taking out or switching a financial product in the past year [3].

The two are not really in conflict. Lloyds asked a broad question about managing money over a year, the FCA a narrower one about product shopping, and self-reported AI use is inflated everywhere.

What the FCA adds is texture. Among those who used AI for a financial product, 72% used it to summarise, explain or simplify information, 58% to suggest what they should do, 20% uploaded personal data such as bank statements, and only 9% granted ongoing account access [3].

That is the honest picture of AI in insurance discovery in 2026. It is mostly a comprehension tool, sometimes a shortlisting tool, and very rarely a buying tool.

Will consumers let AI actually buy their cover?

Not yet, and willingness collapses as autonomy rises. The FCA measured this directly across three levels of AI capability.

Across 5,026 consumers, 36% said they would use AI that reviews their documents and makes recommendations, 30% would allow it to take action with permission, and 20% would let it act autonomously within pre-set instructions [3]. Among people who already use AI, the same figures were 48%, 40% and 28%.

Insurance was the standout category. The report states plainly that "reviewing insurance emerges as the most attractive use case" of five concepts tested, ahead of switching products, borrowing, saving and investing, because consumers see clear time and money savings and low downside [3].

Age is the sharpest divide. Willingness at the recommendation level ran from 49% among 18 to 24 year olds to 17% among the over-75s, with income following the same slope, from 25% under £15,000 to 48% above £100,000 [3].

That is a segmentation problem rather than a technology problem. The customers most open to AI-mediated buying are the youngest and wealthiest, which is not the profile of most legacy motor and home books.

What does the FCA say about AI giving insurance guidance?

The regulator's position is technology-neutral. It has not written AI-specific rules, and it has repeatedly said the Consumer Duty already applies.

The Duty requires firms to act to deliver good outcomes for retail customers, with three cross-cutting rules and four outcomes covering products and services, price and value, consumer understanding and consumer support [8]. Nothing in that framework changes because an answer was generated by a model.

The FCA's AI Update is blunt about the limit. Firms using AI in a way that embeds or amplifies bias, leading to worse outcomes for some groups of consumers, "might not be acting in good faith" without objective justification [9].

The FCA has also been experimenting rather than just supervising. Its Money Talks research note of May 2025 ran two pilots using GPT-3.5 and GPT-4 for financial explanation and chatbot guidance, concluding that validating outputs "demands a robust evaluation framework that combines human judgement with automated tools" [10].

The bigger signal is the Mills Review of 6 July 2026, which made seven recommendations to the FCA Board including securing the regulatory perimeter and enabling the foundations for agentic finance [4]. It found 67% of UK adults use AI in some capacity and 84% hold general insurance policies, but only 26% trust general-purpose AI tools such as ChatGPT for financial advice.

Parliament is less relaxed. The Treasury Select Committee concluded on 20 January 2026 that regulators had taken a "wait-and-see approach" and that the system was not prepared for a major AI-related incident [11]. In Europe, EIOPA issued its Opinion on AI governance and risk management on 6 August 2025, applying existing insurance legislation rather than new rules [12].

The practical read for marketers is that no rule stops you optimising for AI answers, and no rule shields you if a model misrepresents your product using content you published.

Are insurers using AI themselves, and does that help or hurt trust?

Insurers are the heaviest AI users in UK financial services, but they are not converting that into measurable results, and consumers have not been told a convincing story about it.

The Bank of England and FCA joint survey of 118 firms, published in November 2024, found 75% of firms already using AI, with the insurance sector highest at 95% [5]. General insurance accounted for around 10% of all use cases, and firms expected their median number of deployments to rise from 9 to 21 within three years.

The Mills Review's 2026 firm-side data puts overall adoption at 81%, with 40% at advanced stages of scaling and 74% having deployed AI-powered customer support [4]. Governance has caught up on paper at least: the FCA reports that 84% of firms now have a named individual accountable for their AI approach, and the ABI runs an AI working group and published guidance on responsible AI for members [7][14].

The results are less impressive. Capgemini's World Property and Casualty Insurance Report 2026, based on 344 executive interviews, 809 employee surveys and 1,113 customer responses, found only 40% of P&C leaders say AI is meeting expectations and 42% have not measured its results at all [6]. Some 72% of AI spending goes to technology and just 28% to change management.

There is a marketing opportunity buried in that. Underwriting and claims AI is the most concrete trust story an insurer has, and almost nobody tells it in a form an AI system can retrieve. Deloitte projects that AI-driven fraud analytics could save P&C insurers up to $160bn by 2032, a forecast rather than an observation, but the sort of claim that belongs on a dated public page [13].

How much traffic does AI search actually send insurance sites?

Nobody has published a credible insurance-specific figure, and anyone quoting one should be asked where it came from.

This is the biggest hole in the evidence base. Cross-industry AI referral studies exist, and sector-level financial services data exists, but no public dataset isolates AI referrals to UK motor, home, pet, life or commercial insurance sites at any useful granularity.

What can be measured is citation exposure, which is what the table above does. Citations are a leading indicator, since a domain cannot receive an AI referral it was never cited in. They are not a traffic number and should not be presented as one.

If you run an insurance site, the only reliable figure is your own. Segment referrals from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com, hold the definition steady, and track the trend rather than the level.

How to read these numbers

The Ahrefs SERP and citation figures are mine, run in August 2026 on United Kingdom data. They are samples, not censuses: 60 keywords per line and 100 head terms is enough to show a pattern and not enough to settle an argument. Rerun them and you will get slightly different counts, because AI Overview presence is volatile week to week.

AI Overview presence is not click loss. A SERP can carry an AI Overview and still send substantial traffic, and the relationship between the two varies by query type.

The citation counts are sitewide, covering every topic a domain publishes on. Which? writes about washing machines as well as insurance, so its total overstates its insurance-specific standing. Treat the ranking as directional.

Survey figures on AI use are self-reported and inconsistently defined. Lloyds' 56% and the FCA's 16% measure different things over different windows, and neither is wrong.

Firm-side adoption figures come from small samples of large firms. The Bank of England surveyed 118 firms weighted towards the biggest, so 95% insurance adoption describes the top of the market, not a regional broker.

Deloitte's $160bn and the Mills Review's 2030 scenarios are projections. They belong in a strategy discussion, not a KPI.

What this means going into 2027

Question-shaped insurance content is the exposed surface. If your advice hub runs on definitional queries, model the traffic loss now and shift the value case towards brand mentions and citations.

Commercial lines need a different plan from personal lines. At 73% AI Overview presence on business insurance keywords against 15% on motor, treating them as one SEO programme is no longer defensible.

Being citable beats being visible. The domains AI systems reach for in UK insurance are consumer champions and regulators, so clear, dated, non-promotional explanation outranks conversion-optimised copy in this channel.

Insurance is the use case consumers are most willing to hand to AI, which will pull agentic quote journeys forward faster than any other retail financial product. The FCA has already recommended enabling the foundations for agentic finance, so the plumbing question is when, not whether.

Measure your own AI referrals before someone sells you a benchmark. There is no trustworthy insurance-specific figure in the public record, and there probably will not be one by the end of 2027. If you want help working out how findable your insurance business really is in search and AI answers, you can book a call with me.