Measurement & Reporting · 9 min read · July 15, 2026

Measuring AI mentions: Which insurance questions really lead to mandates

Not every AI mention is equally valuable. Whoever surfaces in ChatGPT for "best insurance broker Munich" may win clicks – but mandates arise from concrete questions about specific lines and situations. This guide shows you how, as a broker, to measure which AI answers mention you, which of those bring paying clients, and which insurance questions you should therefore address first.

Why AI visibility works differently for brokers than Google

On Google the prospect clicks on ten blue links and compares for themselves. In ChatGPT, Gemini or Perplexity they get a finished answer with one to three recommended contacts. That changes your business fundamentally: if the AI names you for "I'm self-employed, which occupational disability insurance fits and who advises on it independently?", that doesn't replace a click, it replaces half an initial consultation. The prospect comes to you pre-qualified, with trust and with a concrete line of business in mind.

That's exactly why being mentioned somewhere isn't enough. A mention for the generic question "What does an insurance broker do?" brings you almost nothing. A mention for "Which broker in Augsburg specializes in medical professions and switching to private health insurance?" brings you a mandate with a high closing probability. Measuring AI visibility means, for you, keeping these two cases cleanly apart and building only the second kind on purpose.

As a broker you have an advantage over the big corporations here. Large insurers answer product questions, yet AI models preferentially recommend independent, specialized advisors for advisory and trust questions. If your website, your specialist articles and your reviews clearly show what you focus on, you become the obvious recommendation for exactly these valuable questions – where the corporation can only name a product.

The difference between reach questions and mandate questions

Divide the questions where you could be mentioned into two buckets. Reach questions are broad and informative: "How does term life insurance work?" or "What does private health insurance cost?" They generate a lot of volume, but the asker is rarely ready to buy. Mandate questions are narrow, situational and local: "I'm 34, a doctor, employed, want to switch to private health insurance – which independent broker in Cologne helps me with that?" Such questions already carry buying intent, line of business and region within them.

For your business almost only the second bucket counts. A client who comes through a mandate question already roughly knows what they need and is concretely looking for an advisor. This is exactly where initial conversations arise that lead to ongoing mandates – whether occupational disability cover for the self-employed, commercial insurance for craft businesses or retirement provision for high earners. So don't measure your total visibility, but your visibility per mandate question.

In practice that means: write down 20 to 30 typical mandate questions of your ideal customers, exactly as a real prospect would type them into ChatGPT. Use their words, not your jargon. "Protection against loss of income" becomes, for the customer, "What happens to my salary if I can no longer work?" This list is the foundation for everything else in this guide.

How to measure concretely where the AI mentions you

The simplest method costs you nothing but time. Take your 20 to 30 mandate questions and ask them one after another in ChatGPT, Gemini and Perplexity. For each question, note three things: Are you named by name? Is a competitor named instead? And which source does the AI cite – your website, a comparison portal, an industry directory? Doing this by hand once a month shows you in black and white your starting position per line and region.

It's important that you test locality. Ask once neutrally and once with your city: "…broker in Regensburg" or "…near me, postal code 93047". AI answers are often locally sensitive, especially for brokers. It can happen that you surface for "occupational disability broker Regensburg" but not for the generic occupational disability question. You need to know exactly this pattern to know where you're already strong and where a gap yawns.

If you want it more systematic, there are GEO tools that query such prompts automatically and repeatedly and give you a mention rate over time. To start, though, a simple spreadsheet is enough: question, model, mentioned yes/no, cited source, date. After two or three months you'll see trends – and whether your measures are working or not.

From mention to mandate: closing the attribution gap

The hardest question is: does an AI mention really lead to a mandate? The tricky part is that ChatGPT and the like rarely leave a clickable reference with clean tracking. The prospect reads your name, types it separately into Google or calls directly. In your analytics that shows up as "direct access" or "brand search" – the actual source, AI, stays invisible. This attribution gap is why many brokers underestimate the value of their AI visibility.

You can close the gap with a simple habit: ask every first contact how they came to you, and actively offer "via ChatGPT / an AI / a chatbot" as an answer option. Enter that into your CRM or contact list. After 30 to 50 inquiries you'll have a reliable sense of what share of your new clients comes via AI recommendations – and which line of business dominates.

Combine this customer feedback with your mention table. If you're strongly mentioned for the commercial occupational disability question and at the same time a strikingly high number of commercial occupational disability inquiries come in with "found via an AI", you have your proof. This linking of measurement above (mention) and measurement below (mandate) is the real art – and the point at which GEO turns from gut feeling into a controllable quantity for you.

Which insurance lines are worth pursuing first for AI visibility

Not every line is equally worthwhile. Prioritize by three criteria: how high is the advisory intensity, how good is the commission or portfolio compensation, and how strongly do people really search a AI for an advisor on it? Occupational disability, private health insurance, commercial and business liability, and company pension schemes are almost always at the top. They need explanation, are long-lived in the portfolio, and people are reluctant to make a purely self-directed decision here.

Motor or household contents insurance, on the other hand, are often price- and portal-driven business. Here the AI recommends comparison calculators rather than a personal broker, and the margin rarely justifies the effort. That doesn't mean you drop these lines – but you invest your limited time for content and optimization first where a single won mandate question pays off for years.

A realistic example: a broker focusing on medical professions deliberately addresses questions like "As a junior doctor, switch to private health insurance or stay in statutory?" and "What occupational disability cover do I need as a female dentist?" This niche is narrow enough that he often surfaces in AI answers as a specialist, and valuable enough that every mandate is worth it. Breadth almost never beats depth here.

What you concretely optimize so the AI mentions you

AI models mention you when three signals line up: unambiguous specialization, verifiable substance and external confirmation. For specialization, write unmistakably on your website who and what you stand for – not "all insurance for everyone", but "independent broker for occupational disability and private health insurance for doctors and the self-employed in Eastern Bavaria". This clarity is what an AI can even reproduce as a recommendation in the first place.

You build substance with specialist content that answers real mandate questions. A cleanly written guide "Occupational disability cover for doctors: the five typical mistakes when signing up" gives the model quotable material and positions you as a source. Pay attention to clear structure, concrete figures and honest assessment – AI systems prefer content that answers a question completely and neutrally over pure advertising copy.

External confirmation, finally, comes from Google reviews, professional directories, press mentions and entries with broker pools or associations. If several independent sources connect you with the same specialization, the probability rises markedly that an AI mentions you for the matching question. Make sure these sources tell the same positioning as your own page consistently.

A simple monthly measuring ritual for your firm

Make the measurement a routine, otherwise it fizzles out. Block off 60 minutes once a month. In the first half hour you ask your fixed list of mandate questions in ChatGPT, Gemini and Perplexity and enter the mentions into your table. In the second half hour you look into your CRM at how many of the new first contacts in the previous month stated "via an AI" and which line they belonged to.

Define two or three metrics that you really track: the mention rate across all mandate questions, the mention rate in your two most important lines, and the share of AI-attributed first contacts. You don't need more than that at the start. More important than perfect numbers is repetition: only the trajectory over several months shows you whether a new guide article or more reviews actually moved your visibility.

If a metric doesn't rise over two months, you change one thing deliberately – sharper specialization on the homepage, an additional specialist article on the affected question, active collection of reviews. That's how AI visibility becomes not a gamble, but a controllable channel that you take as seriously as referrals or your portfolio work.

Honest limits: what measuring doesn't promise you

Be realistic: AI answers fluctuate. The same question can deliver your name today and a competitor's tomorrow, depending on model version, wording and chance. Your measurement is therefore a sample, not an exact ranking. Only draw conclusions from patterns over several weeks, not from a single answer that thrills or annoys you.

Equally honest: AI visibility replaces neither good advice nor your referral business. It's an additional inflow channel that often brings especially pre-qualified prospects, but rarely delivers masses in the short term. Anyone who starts today, measures and patiently works on specialization and substance builds up over six to twelve months a lead that late-starting competitors struggle to close.

And finally: sales-law rules still apply. Your content must be correct and balanced, your reviews genuine, your advice documented. GEO is not a trick to circumvent rules, but the consistent digital translation of what distinguishes you anyway: clear specialization, reliable substance and a good reputation that a machine can now read too.

Common questions

I'm an all-round broker without a clear niche. Is AI visibility even worth it for me?

Yes, but you first have to choose one or two focus lines in which you want to be visible – say occupational disability cover for the self-employed or commercial insurance for craft businesses. AI models preferentially recommend recognizable specialists. Feel free to remain an all-rounder outwardly, but build your AI visibility deliberately around your most profitable, advisory-intensive lines. Breadth almost never beats clear depth in AI recommendations.

How do I recognize whether a new client really came to me via ChatGPT?

Don't rely on your website analytics; there AI traffic usually shows up only as direct access or brand search. Instead, actively ask every first contact how they found you, and explicitly mention the AI or chatbot option. Enter the answer into the CRM. After 30 to 50 inquiries you'll have a reliable rate of what share of your mandates arises via AI recommendations.

Which insurance line should I optimize first for AI visibility?

Start with a line that has high advisory intensity, good portfolio value and genuine search intent for an advisor – typically occupational disability, private health insurance, commercial/business liability or company pension schemes. Motor and household contents are portal- and price-driven business; there the AI recommends comparison calculators rather than you. Invest your limited time first where a single won mandate carries earnings for years.

Read on

Brand & Positioning