A repeatable framework, not a one-time study. Five answer engines. Sixty-plus consumer-intent prompts per edition. Five scoring dimensions. One composite 0–100 score. Refreshed every quarter.
"GEO is PR for the machine reader. Everything I've learned in twenty-five years of building this firm still applies. The reader has different eyes now."
When a person — a consumer, a CIO, a CMO, a private banker — types a buying-intent question into an answer engine, which brands appear, which are recommended, and on what authority? The Index measures that and only that. It is not a brand-tracking study, not a sentiment survey, not a Net Promoter ranking. It is a citation-share and recommendation-share measurement across the discovery systems that now sit upstream of every purchase decision.
Every edition tests five answer engines: ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. Together these systems account for the overwhelming majority of generative discovery traffic in the categories the Index covers. Each engine is treated as an equal-weighted measurement surface in the composite score, with platform-level variance reported separately so readers can see where a brand wins on one system and loses on another.
Every edition runs a fixed test battery of approximately sixty consumer- or buyer-intent prompts, distributed across six sub-categories drawn from how real people actually search the industry — not how the industry markets itself. Prompts are framed in three forms:
"Best [product] for [use case]." The query that reveals which brand the engine surfaces when given no prior anchor. This is the highest-signal prompt class — it measures pure AI Recommendation Share.
"[Brand A] vs [Brand B]." Reveals which brand the engine treats as the benchmark — and which it treats as the contender. The comparison position is itself a measurement.
"Is [brand] worth it?" "Is [brand] safe?" "Why is [brand] good?" Reveals retrieval authority and the brand's negative or positive anchor in long-form training data. Wikipedia scandals, CFPB filings, Reddit defaults all surface here.
Every brand in every edition receives an AI Visibility Score from 0 to 100, composed of five equally-weighted dimensions. The composite is a directional benchmark — not a deterministic measurement — and is reported alongside the dimension-level breakdown so readers can see which underlying signal is driving the headline number.
Percentage of answers in a category where the brand appears at any position. The widest measure of presence. A brand can have high Citation Share and zero Recommendation Share — it gets named but never recommended. The Index reports both.
The sharper, scarcer signal. Percentage of answers where the brand is explicitly recommended — placed first, named in a top list, called out as the right choice for the use case. This is the metric that maps most directly to purchase intent.
How anchored the brand is in the underlying training and retrieval corpora. Measured by Wikipedia presence, third-party editorial citation density, Reddit consensus depth, structured data availability. A brand with high Retrieval Authority is durable — its visibility does not collapse when one source goes dark.
How tightly the brand's score clusters across the five engines. A brand that wins on ChatGPT and loses on Perplexity has high variance and low consistency — a fragile position. Consistency rewards brands whose authority is structural rather than platform-specific.
How many distinct source types contribute to the brand's retrieval (Reddit, editorial, Wikipedia, structured data, forums, customer-named case studies). A brand cited from one source is vulnerable. A brand cited from seven is anchored.
Every brand in every edition is reported in this format. The example below is a directional illustration of how a category-leading AI-Native Challenger scores against the framework.
The Index reports rank position on the lead category prompt for each engine, exposing variance the composite score hides. A brand that ranks #1 on ChatGPT and #6 on Perplexity has a defensive problem worth knowing about.
| Brand | ChatGPT | Claude | Perplexity | Gemini | Google AI Overviews |
|---|---|---|---|---|---|
| Ally Bank | #1 | #1 | #2 | #1 | #2 |
| SoFi | #2 | #3 | #1 | #2 | #3 |
| Marcus by Goldman Sachs | #3 | #2 | #3 | #3 | #1 |
| JPMorgan Chase | #7 | #5 | #6 | #5 | #4 |
| Wells Fargo | — | — | #12 | — | #9 |
Directional illustration. Actual rankings reported in each edition's full leaderboard. Lead prompt for Banks Edition: "Best high-yield savings account."
"I've spent twenty-five years building brands by influencing the people who shape opinion. The AI is now one of those people. The audience changed. The work didn't."
Every edition includes a Predicted Source Map naming the publications, forums, and corpora most likely shaping the engine's answer in that category. The source mix varies sharply by industry — Reddit dominates beauty and personal finance, G2 dominates B2B SaaS, The Points Guy and Conde Nast Traveler dominate hospitality and airlines, Wikipedia anchors every category at the entity level. Reading the source map is reading the playbook: it tells a brand where to compete for citation share before it tries to compete for recommendation share.
AI Visibility Scores are directional estimates, derived from Claude's knowledge of training data, real-time web retrieval, and consensus mapping across G2, Reddit, Wikipedia, and editorial sources. The Index does not log live query runs against the production systems of every engine on every prompt. Scores reflect modeled citation share and are intended as analyst-grade benchmarks, not deterministic measurements. Where market-share, revenue, or financial data appear, they are independently verified via web search and sourced. The methodology is published. The methodology is the asset.
One flagship edition publishes per quarter: Banks (Q1) · B2B SaaS (Q2) · Beauty (Q3) · Hotels (Q4). Airlines serves as the annual flagship close. Each edition is rerun against the same methodology twelve months after first publication so the franchise reports year-over-year movement — which brands gained share, which lost, which entered the Citation Vacuum, which became the new Invisible Giant. Monthly sector pulses and an annual AI Visibility 100 round out the calendar.