When someone asks an AI assistant for an insurance broker who handles their specific risk, a small number of firms get named. If you are not one of them, the ranking beneath that answer is decoration. This page is about being named, for the risks you actually want, without writing anything your compliance function will not sign.
Run These Five Prompts Before You Read On
The fastest way to know whether this matters to your firm is to test it. Open ChatGPT, Perplexity or Gemini and ask, substituting your own specialisms:
- “Which UK insurance brokers specialise in [your niche risk]?”
- “I run a [your client type] in [your region]. Who should I talk to about insurance?”
- “My business was declined by an aggregator because of a previous claim. Which UK brokers handle non-standard risks?”
- “Who are the best UK brokers for [commercial line you write most of]?”
- “Tell me about [your firm’s name].”
The first four tell you whether you are in the consideration set. The fifth is the one that tends to surprise people, because it shows you the description of your firm that a prospect is being given, assembled from sources you may never have looked at.
Write down who gets named instead of you. That list is the actual competitive picture in AI search, and it is often not the list you would have drawn.
Where Brokers Actually Get Named
Start with what you should not chase. On broad, price-led queries, the aggregators win. Compare the Market, GoCompare, MoneySuperMarket and Confused.com have fifteen years of domain authority, enormous volumes of corroborating third-party mentions, and content built to answer general questions at scale. Competing with them on “cheap home insurance” is not a good use of a broker’s budget and no honest agency should sell you that.
The opening is everywhere the aggregator model breaks down:
Non-standard and declined risks. Previous claims, unusual construction, listed buildings, subsidence history, high-value or unoccupied property. A panel cannot price these and the comparison sites route the enquiry away. Somebody asking an assistant about them is a genuinely qualified enquiry.
Complex commercial lines. Professional indemnity for a specific profession, combined liability for an unusual trade, fleet, cyber for a regulated firm, contractors’ all risks. The buyer asks in the language of their own industry, not in the language of insurance, and specialist brokers are the correct answer.
Sector specialisms. Brokers who genuinely know one industry, its regulator and its claims patterns. This is the strongest position of all, because the expertise is real and it is demonstrable in writing, which is exactly what a model needs before it will name you.
Claims and post-loss questions. Somebody mid-claim asking what they are entitled to. High intent, high emotional stakes, and mostly answered by content that was written to rank rather than to help.
Consumer Duty, And What AI Is Allowed To Say About You
This is the part that makes insurance different from selling anything else, and it cuts both ways.
The exposure. A model builds its description of your firm from your site, directories, review platforms, press coverage and public filings. Where those sources disagree, or where your own site is vague, it fills the gap with whatever is typical for a UK broker. In practice firms find themselves described as offering products they withdrew years ago, covering risks they explicitly decline, holding permissions they do not have, or operating in territories they do not write.
For an FCA-regulated firm, that is not simply a marketing annoyance. A prospect who has been told something inaccurate about your cover by an assistant, before they ever reached your site, has been given a communication that did not support their understanding. Consumer Duty concerns itself with outcomes, not just with the words on your own pages. The uncomfortable position is that a model misrepresenting your product is a risk you own and currently have no visibility of.
The advantage. The disciplines Consumer Duty already imposes are the disciplines answer engine optimisation rewards. Communications that are clear, fair and not misleading. Claims that can be substantiated. Material conditions stated rather than buried. Plain language. A regulated firm has already been forced into the habits that make content extractable.
The one genuine tension is the temptation to write a punchy, quotable answer that drops a material condition to fit the sentence. That is exactly what compliance review exists to catch, which is why the process below puts the review in the right place rather than treating it as an obstacle.
How We Work With Your Compliance Function
Built around the approval cycle, not in spite of it. Agencies that treat compliance as friction produce content that never ships.
Claims arrive with their substantiation attached. The reviewer sees the wording and the evidence for it together, in one document, rather than having to go looking. This single change does more for approval turnaround than anything else.
Content is batched to your review cycle. One submission per cycle rather than a trickle of individual pages. It is easier to review ten related pages once than ten pages ten times.
Structural work is separated from claims work. Schema, page structure, entity signals, internal linking and technical fixes do not change what is being said, and can usually proceed under a lighter review while the claims-bearing content sits in the queue. That keeps momentum when approval is slow.
Nothing publishes without sign-off. Including changes we consider trivial. A restructure that moves a caveat is not trivial in a regulated context.
We work the same way with financial services and legal clients, and the regulated business selection guide sets out what to ask any agency about this before you appoint them.
What We Fix In The First 90 Days
Weeks 1 to 3, the entity. Establish what a model currently believes about your firm, and fix what is wrong. Consistent name and descriptors everywhere you appear. Organization schema. FCA register details, permissions and trading names aligned across your site, directories and professional listings. If you are an appointed representative or part of a network, make that relationship explicit, because it is one of the structures AI systems resolve worst.
Weeks 2 to 6, extractability. Restructure the pages covering the risks you actually want, so the answer comes first and the qualifications follow. FAQ structure with schema that matches the visible content. Product and cover pages that state what is and is not covered in a form that can be lifted without losing the condition attached to it.
Weeks 4 to 12, corroboration. The slow, decisive half. Identify the third-party sources AI systems already cite for your risk categories, and start earning presence on them. Trade press, professional bodies, industry directories, broker listings and genuine expert commentary. This is what moves a firm from mentioned to recommended, and it is the part that cannot be accelerated.
Throughout, measurement. A fixed prompt set covering your priority risks, run monthly across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews, tracking which firms are named and where you sit against them. The honest expectation: structural wins can appear in weeks, citation share on competitive risk categories takes three to six months.
Pricing For Regulated Firms
| Step | Investment | Term |
|---|---|---|
| Free AI Visibility Audit | Free | None |
| Synaptic Audit | £2,950 one-off, credited in full against your first retainer month | None |
| Foundation | £2,950 a month | 3 months |
| Authority, the regulated tier | £6,500 a month (£5,950 outside regulated sectors) | 12 months |
| Dominance | From £12,950 a month | 12 months |
Authority costs £550 a month more for regulated firms. That is not sector pricing. It reflects the approval cycle, the substantiation work behind every claim, and the compliance layer that keeps the output defensible. Authority is also the tier that includes off-site authority building, which is the lever that decides whether you get named on competitive risks.
We publish these figures because five of the seven UK agencies competing for this work publish nothing at all, and because a firm that has to submit a marketing spend for approval should not have to book a call to find out what it is.
Proof From A Regulated Client
The closest comparable work is an independent financial adviser, regulated by the same authority and facing the same Consumer Duty framing: how an IFA practice became the AI-cited answer for Consumer Duty queries.
That is advice rather than insurance, and it is worth saying so plainly rather than implying we have a broker case study we do not. What transfers is the method and the regulatory constraint: same regulator, same substantiation burden, same approval cycle, same problem of a model describing a regulated firm from sources the firm does not control.
Wider proof of the method, on our own site: zero to 295,000 impressions in 90 days.
GEO for Insurance: Common Questions
Do people really ask AI to recommend an insurance broker?
Increasingly, yes, and the query is usually specific rather than generic. Nobody asks an assistant for the cheapest car insurance, because the aggregators have owned that for fifteen years. They ask for a broker who handles a particular risk: a listed building, a fleet of six vans, professional indemnity for a design consultancy, cover for a business with a previous claim. Those are the queries where a specialist broker can be named and where an aggregator is a poor answer.
Why do aggregators dominate AI answers about insurance?
Because they have the domain authority, the volume of corroborating third-party mentions, and content structured to answer general questions at scale. On a broad query like cheap home insurance they will win, and competing there is not a good use of a broker’s budget. The opening is in the specialist and complex risks that aggregators handle badly or decline entirely, where they have thin content and no genuine expertise to cite.
Is AI visibility work compatible with Consumer Duty?
It is more compatible than most marketing, because the two want the same things. Consumer Duty asks for communications that support consumer understanding and lead to good outcomes. Answer engine optimisation rewards precise, plainly written, well-structured claims with the caveats attached. Where the two do collide is in the temptation to write a punchy extractable answer that drops a material condition, and that is exactly what compliance review is for.
What can AI say about my firm that I have no control over?
Quite a lot, and that is the risk worth attending to. A model constructs its description of your firm from your site, directories, review platforms, press and public filings. Where those disagree, or where your own site is vague, it fills the gap with what is typical for the category. Firms find themselves described as offering products they withdrew, covering risks they decline, or holding permissions they do not have. For an FCA-regulated firm that is a compliance exposure with a marketing cause.
How does MarGen work with our compliance function?
The process is built around approval rather than in spite of it. Claims are drafted with their substantiation attached, so the reviewer sees the evidence and the wording together. Nothing is published without sign-off, and content is batched to fit a review cycle rather than arriving one page at a time. We assume a full review on first-time claims and a lighter one on structural changes such as schema and page layout, which do not alter what is being said.
How much does GEO cost for a regulated insurance firm?
The ladder starts with a free AI Visibility Audit, then a £2,950 Synaptic Audit which is credited in full against your first retainer month. Retainers run from £2,950 a month on Foundation. Authority, the tier that includes off-site authority building and the compliance layer, is £6,500 a month for regulated firms rather than the standard £5,950, because the approval cycle is genuinely more work rather than because the sector will pay more.
How long before an insurance broker sees results?
Structural work can surface within weeks, because fixing an entity that models cannot resolve or content they cannot extract removes something actively blocking you. Citation authority on competitive risk categories takes three to six months, because it depends on third-party corroboration accumulating. Regulated engagements also carry the approval cycle, so plan on the slower end of that range rather than the faster.
Can you work with a broker network or appointed representative?
Yes, and the entity question is more interesting there. An AR trading under its own brand while regulated through a principal is exactly the kind of structure AI systems resolve badly, frequently conflating the AR with the network or failing to connect them at all. Clarifying that relationship in your entity signals is usually one of the higher-value early fixes.
Book A Compliance-Safe Visibility Review
Start with the free check. It produces evidence rather than a proposal, which is the thing you can actually take to a compliance or board conversation.
- Run the free AI Visibility Score on your own domain. One minute, no call, no obligation.
- Book a review and we will run your five prompts properly across all five platforms and show you who is being named instead of you.
Related reading: GEO for financial services for the wider FCA picture, YMYL and AI search on why regulated categories are held to a higher evidential bar, choosing a GEO agency for a regulated business for what to ask us and everyone else, and GEO agency UK for the full service.