Insurance shopping moved into the conversation.

It is restricted rather than banned, and the position keeps moving. Financial services is listed among OpenAI's restricted advertising categories, while OpenAI stopped categorically blocking ads near financial advice in April and eligibility has changed four times since March. We confirm your specific position in writing and build the rest of the programme so nothing depends on one approval.

Because the products are genuinely hard to compare and the questions are long. That is the exact job people now delegate. 18.7% of home insurance shoppers and 13% of motor shoppers already use AI tools, ahead of brokers at 14.5% in home, and that reading came from the very first quarter the channel was tracked.

Cost and coverage answers with real numbers. Pricing drove 17.6% of insurance citations, the leading intent, and generic explainers lose to pages that state figures, conditions and exclusions plainly. It takes compliance involvement, which is precisely why so few carriers do it and why the opening is still there.

Yes, and this is the most encouraging finding in the data: 53.3% of insurance AI Overview citations went to pages that never cracked Google's organic top 10. Assistants reward specificity and clarity over domain authority, so a focused broker or insurtech can be cited beside national carriers.

They are in the process from the start, not asked to approve at the end. We agree what can be said about pricing, coverage and comparisons before writing, submit content in batches, and keep a record of approved claims so the programme scales without a bottleneck. Nothing publishes without your sign-off.

No. Search still carries the majority of quote demand and should keep doing so. What changes is that some comparison now happens before the search ever occurs, so a share of budget belongs where that conversation happens. With a billion weekly ChatGPT users, 20% of them showing commercial intent, it is worth a deliberate test rather than a reallocation.

Partly with data and partly by asking. Referral data where the platform passes it, call measurement on agent lines, quote-start and bind events, and a source question in the flow. It matters that we are honest about the limits: 59% of insurance consumers cannot reliably tell AI from a person, so self-reported source data is directional, and we label it that way.

More than most marketing plans assume. 33% of Boomers have used generative AI to research a high-stakes insurance purchase, up from 11% a year earlier. Growth is fastest among exactly the segment carriers assumed would be last, which changes both the media plan and the tone of the content.

All the ones your buyers use, and the citation split is not what people expect: Google AI Mode drove 43.9% of analysed insurance citations with Perplexity at 33.6%. The content and structured data work serves all of them at once, which is why we treat it as the backbone rather than a ChatGPT-specific tactic.

Weighted towards content and measurement in the first quarter, with a contained paid test once eligibility is confirmed. The citation work keeps paying regardless of platform policy; the ad spend is the part that can be switched off overnight by a rule change. We would rather build the durable half first and tell you so.

They change what a broker is for rather than removing the need. The assistant handles the first comparison, so the broker's value moves towards complex risks, claims support and advice a model will not give. Brokers who publish genuinely useful coverage guidance get cited by the same assistant and picked up demand from it.

Paid placement, once approved, produces quote starts within days. Citation share typically shifts over two to four months as content is indexed and picked up. We report prompt testing monthly so you can see movement long before it reaches written premium, and we say when a month showed nothing.

We write it, from interviews with your product, underwriting and claims people, then your compliance team reviews before anything publishes. Insurance content written without those conversations reads generic, and generic is exactly what assistants skip when choosing a source to cite.

Nothing in this programme requires sharing customer data with an ad platform. Measurement is built from your own analytics, call records and quote events, with consent handling reviewed alongside our privacy compliance team. Where a tactic would create a data risk, we leave it out and explain why.

Ask what they measure and what they promise. Guaranteed placement inside assistant answers is not a product anyone can sell. What you should expect is a recorded prompt baseline, a written eligibility position, compliance-ready content and reporting that ends at bound policies rather than impressions.

It runs inside it rather than beside it. The same optimization work — clear pages, structured product data, credible third-party coverage — feeds both organic rankings and AI citations, so we plan it with our SEO team and report the two together. Brands with a healthy organic foundation reach citation share faster and spend less doing it.

Usually auto and home, because that is where AI shopping behaviour is furthest along and where quote volume makes a test readable within a quarter. Life and specialty lines carry longer consideration and suit the content side first. We pick the starting line from your quote-to-bind data and your marketing capacity, then widen once one line shows a repeatable lead and conversion pattern.

Prompt baseline across the assistants, written eligibility and compliance read, content plan ranked by citation opportunity, and a measurement review of your quote funnel. You finish the month knowing exactly where you are cited today and what the next quarter is meant to change.