A playbook for measuring brand visibility in AI-generated answers

Incumbents capture over 78% of brand mentions in AI answers. A systematic playbook from Geology AI shows how challengers can measure their visibility gap before rewriting a single landing page.

Incumbents take 78 to 82 percent of insurer mentions on ChatGPT, Perplexity, and Gemini, according to a June 2026 study by the firm Geology AI. For challenger brands, this means AI answer engines are effectively a blackout. The typical response is to rewrite landing pages, a guess aimed at a black box. The Geology AI post argues this is the wrong first move. This is a measurement problem before it is a content problem. You cannot improve what you do not track.

The landscape: aggregators and niches

The Geology AI study analyzed 100 insurance-related prompts across four major AI platforms. It claims that for most general queries, challengers are invisible. The engines overwhelmingly cite external aggregators and editorial sites like Forbes and NerdWallet, not the brands themselves. On ChatGPT, the study reports only about 1 percent of sources were carrier-owned websites.

The data also showed where challengers can win. On Perplexity, Reddit was the single most-cited source, appearing in 75 of 100 answers. This suggests a surface where authentic discussion can outperform corporate content. The study also noted that the few visible challengers, like Next Insurance and Lemonade, won by owning specific niches like small business or renters insurance, not by competing on general terms.

Step 1: build a repeatable prompt set

The playbook begins with building a test harness. Instead of ad-hoc checks, Geology AI prescribes creating a structured set of 40 to 100 prompts. These prompts should be stored as data, like a JSON file, to make testing repeatable. This is a critical step for tracking visibility over time as models and content change.

The prompts are sorted into four distinct buckets to mirror user search patterns:

• Category: Broad queries like "best renters insurance 2026".
• Problem: Specific user needs like "how do I insure a side business".
• Comparison: Head-to-head queries such as "Lemonade vs State Farm renters".
• Niche: The specific use cases where a challenger brand believes it has a right to win, for example "renters insurance for a freelance photographer with gear".

The source document from Geology AI is incomplete. It cuts off after describing this first step, before detailing the full measurement process.

What we'd change

The playbook's core premise is sound. Treat AI visibility as a data problem, not a content-guessing game. But its application requires context. The provided steps are tailored to a high-volume consumer category like insurance, where aggregators and media sites have already built deep content moats. For a B2B SaaS tool in a new category, the competitive landscape and influential sources will be entirely different. Reddit might be irrelevant, while specific industry blogs could be the primary sources for AI models.

The methodology also implies a technical founder. Building a JSON prompt set and presumably scripting the process of running them against multiple AI APIs is not a marketing task for most teams. A non-technical founder would need to adapt this into a manual, but still structured, process using a spreadsheet.

Finally, the incomplete nature of the source material leaves the most important part unanswered: what to do with the measurements. The logical next step is a targeted campaign to appear on the sources the AI engines already trust. This means shifting effort from direct-to-consumer SEO to getting mentioned on NerdWallet, seeding discussions on Reddit, or placing content in Forbes. The goal is to be the signal, not just the destination.

Landing

For founders outside the top 1% of venture-backed unicorns, direct competition is a losing strategy. The same is true in AI answer engines. Incumbents and major publishers have an insurmountable advantage in training data. This playbook, though incomplete, provides the correct starting point. It reframes the challenge from shouting into the void to finding the specific, narrow cracks where a new brand can be the most relevant answer. Measurement comes first.

The investor read

The emergence of "Answer Engine Optimization" (AEO) creates a new diligence category for investors. A startup's visibility within AI-generated answers is a leading indicator of its future organic traffic. The 78-82% incumbent dominance reported by Geology AI in insurance is a useful, if unverified, benchmark for other mature categories. Investors should ask portfolio companies how they measure their "share of answer" for core buying intent queries. A team that can present a systematic measurement process, like the one outlined here, signals a higher level of marketing sophistication than one focused purely on traditional SEO. The strategy also shifts. Capital is better deployed getting a brand mentioned on high-authority aggregators and forums like Reddit, which AI models cite, than on optimizing the brand's own domain.

Pull quote: “This is a measurement problem before it is a content problem.”