.png)
Once upon a last quarter, a marketer walked into the woods.
She had gold stars on her dashboard. Rankings. A respectable bounce rate. The kind of CTR graph that makes a VP of marketing calm down for an entire afternoon.
And yet.
A nagging feeling. A cold wind rustling through the Slack channel. The board kept asking whether her brand was winning in AI, and every time she tried to answer the question, the answer felt made up.
Is anyone seeing us in ChatGPT? Is Perplexity citing us? Does Claude recommend us or quietly shove us to the back of the list like an uninvited plus one? Nobody knew. Her agency didn’t know. Her competitor’s LinkedIn post claimed they knew, but that post was also an ad for a webinar.
So our heroine, in the grand tradition of women with specific taste and three chairs to sit in, decided to go find a measurement framework that actually fit.
She found three.
Reader, they were not all the same.
Bowl One: Too Cold
The “ask ChatGPT once and call it strategy” approach.
Goldilocks sat down at the first table. The bowl was small. The spoon was a vibe.
This is the measurement style of the brand that opens a tab, types “what are the best [category] brands,” squints at the result, and reports back to the team that things are “generally fine” or “kind of bad” based on what they saw that one Tuesday afternoon.
It is charming. It is fast. It is also, statistically, a lie.
AI responses are probabilistic, not deterministic. Search the same keyword twice and Google mostly produces the same podium. AI works differently. When a user asks ChatGPT, Perplexity, or Google's AI Mode a question, the system assembles a response from dozens of sources and decides, in that moment, which brands to name, cite, and recommend. Run it again an hour later and the cast may have rotated.
Early research into AI citation consistency suggests that only a fraction of brands stay visible from one AI answer to the next, with even fewer holding their position across multiple consecutive runs. So when a marketer tests a prompt once and reports back, they're quoting the weather from a single moment and calling it the climate.
Rankings were the scoreboard. One prompt is just a screenshot.
Goldilocks put the spoon down. She had a suspicion this bowl was not actually measurement. It was a feeling wearing a spreadsheet.
How do you measure AI visibility?
Bowl Two: Too Hot
The “my SEO dashboard already covers this” approach.
The second bowl was steaming. Dense. Important looking. It had seventeen tabs open and a BI certificate on the wall.
This is the measurement style of the marketer who believes that because they are tracking everything, they must be learning something. Organic impressions. Keyword positions. Backlink velocity. Domain rating. A custom dashboard with a feature called “AI Signals” that is actually just a chart of branded search volume overlaid on a rainbow.
The problem is not the effort. The problem is the instrument.
Rankings measure position in a deterministic list. A podium. AI visibility measures presence in a generative narrative that changes every time someone hits enter. A brand can sit at position one on Google and be completely absent from the AI answer for the exact same query.
eMarketer confirms the gap. Fewer than 9% of ChatGPT and other AI citations come from top-10 Google URLs. Which means the typical SEO dashboard has visibility into less than a tenth of the sources AI is actually using to describe your brand. Any measurement framework that conflates ranking with being cited undercounts the problem by more than 90%.
The instrument is pointed in the wrong direction.
That does not mean SEO is dead. SE Ranking’s study of 2.3 million pages found that domain traffic is the top predictor of AI citations. SEO is the foundation. It is also no longer the final scoreboard.
Goldilocks closed the laptop. Too much data. Too little signal. She had been optimizing a playing field that wasn’t the one the game was being played on.
Bowl Three: Just Right
The nine metrics that actually tell you what AI thinks of your brand.
The third bowl was not dramatic. It did not have a brand color. It was simply the right size and temperature for the question at hand.
This is where a new category of measurement lives. It is called generative engine optimization, and the metrics it uses are specifically designed for a world where the answer is generated instead of ranked.
Gist GEO tracks nine baseline metrics. Each one answers a different question a marketer used to be able to answer by glancing at a position tracker.
1. Share of Voice. How often does AI name you?
Share of Voice answers the most basic question in the stack: When the AI is talking about your category, how often does it say your name out loud? If 100 prompts return 34 mentions of your brand and 52 of a competitor, the gap is no longer a vibe. It is a number. Numbers can be moved.
This is the starting line. Without it, you are adjusting your aim in the dark and hoping the target moves toward you.
2. Sentiment. How does AI talk about you when it talks about you?
Being named is good. Being named warmly is better. Sentiment measures the tone AI uses when it describes your brand, because there is a meaningful difference between “a solid, well-reviewed option” and “a brand some users have complained about.” Yes, AI has a tone. Yes, it can be measured. No, you probably do not want to see the raw data before coffee.
3. Share of Citations. Does AI trust your content enough to link to it?
Mention is a guest list. Citation is an invitation with your name spelled correctly. A brand can have a strong Share of Voice and a weak Share of Citations, which translates to: the AI talks about you at the party but never introduces you to anyone. Your PR team will recognize the feeling.
4. Share of Found Links. Can AI actually reach the pages it wants to cite?
A citation only counts if the AI can get to the page. Share of Found Links measures whether your citable content is reachable and indexable when AI goes looking for it. A great page that a crawler cannot open is a billboard in a basement.
5. Earned Media Score. Is the rest of the internet vouching for you?
Earned Media Score measures the strength of third-party mentions AI leans on to describe your brand. It asks whether publications, forums, and reviewers are talking about you in the places AI is listening. Because AI is listening.
6. Citation Rate. When AI does its homework, how often are you in the reading pile?
Citation Rate measures how often your pages show up in the research set an AI assembles for a given answer, whether or not it quotes you in the final text. Think of it as the difference between “we got quoted” and “we were on the syllabus.”
7. Share of Recommendations. Does AI actually endorse you?
There is a serious difference between being listed alongside ten vendors and being the one the AI actually recommends. Semrush reports that AI-driven visitors convert at 4.4x the rate of organic visitors. The reason is not mysterious. By the time the visitor clicks through, the AI has already done the pre-qualifying. The sales team just has to show up and not ruin it.
8. Placement. How early in the answer does your name appear?
Position still matters. Anyone who has ever sat at a long dinner table knows that the person in seat one is part of the conversation, and the person in seat nine is waiting to be asked where they are from. In an AI response, the brand mentioned first shapes the reader’s decision. A brand named last, after three competitors, is technically visible and functionally background music.
9. Average Rank in Lists. When AI ranks the field, where do you land?
When an AI produces an ordered list of options, Average Rank in Lists tracks where your brand tends to sit across many answers. Showing up is one thing. Showing up near the top, consistently, is the thing that moves consideration.
How the nine metrics map to Brand Health
Together, these nine metrics produce a dashboard that shows how strongly your brand is showing up and how much of an impact it is making while it is there.
The part where Goldilocks actually solves the problem.
Individual metrics are a start. Brand health is the whole meal.
Looking at Share of Voice alone is like checking one stat of a basketball game and deciding who won. Gist GEO rolls all nine baseline metrics into four brand health dimensions. Each is scored 0 to 100. Each is updated weekly. Together, they produce a composite picture of how the world sees your brand.
Awareness and Sentiment
Awareness measures whether people and AI systems know your brand exists. It is determined by Share of Voice from GEO data and Source of Authority mentions from Gist Signals.
Sentiment measures how people feel about the brand. It is determined from review sites, social listening, and AI response tone analysis. Yes, AI has a tone. Yes, it can be measured. No, you probably do not want to see the raw data before coffee.
Authority and Recall
Authority measures whether your brand is cited as a trusted source, determined by Share of Citations, Share of Found Links, Earned Media Score, and Citation Rate.
Recall measures whether AI and people can name your brand when directly asked, determined by Share of Recommendations, Placement, and Average Rank in Lists.
Each of these four dimensions can hold their own in a pinch. They are only truly strong when they stand together.
Brand Health inside the seven-dimension Health Framework
Brand Health is one of seven dimensions in the Gist GEO Health Framework. The other six: Product Health, Agent Health, Velocity, Connections Health, Signals Health, and Context Health. They measure everything from whether users love the product to whether AI can access pages on your site. Together, all seven compose a single Growth Score, the overall health of a brand’s growth operation.
There is no longer one single number that captures brand visibility in AI the way rankings used to for search. That is what the Health Framework is for. It helps you figure out what is going on, decide what matters most, and actually take action.
Goldilocks, for what it is worth, liked this part best. It was the first time the data felt like a map instead of a pile of clippings.
The measurement cycle. Or: how Goldilocks learns to stop panicking and check the porridge weekly.
Why one prompt is not measurement.
Asking ChatGPT a handful of questions, glancing at the answers, and calling it a day is not measurement. AI responses vary by session, model, and user context. A single prompt tested once tells you almost nothing. Like the weather, AI responses shift on a regular basis.
Proper measurement requires a representative prompt library (50 to 100 queries across awareness, consideration, and decision stages), consistent testing across multiple AI platforms, and repeated runs to account for volatility. You are not looking at snapshots. You are tracking patterns over time.
PAUSE. TASTE THE RIGHT BOWL. Pick one AI platform. Ask three questions a real buyer would ask in your category. Track two things: who gets named, and who gets named first. Then read your horoscope. If you were named in all three: that is your Share of Voice baseline. Now do it across multiple AI models and run it weekly. Congratulations, you have started measuring. If you were named first in at least one: that is a signal of Placement. You have influence over the reader’s first impression. Guard it. If you were named but never recommended, or recommended but never cited with a link: you have a lopsided Brand Health score waiting to be diagnosed. If a competitor was the only name mentioned across all three: print this page. Walk it to your CMO. Circle the nine metrics.
The four-stage measurement cycle
Measurement is not a one-night stand. It is a standing appointment.
If you want your brand to stay in shape, you have to keep hitting the gym. eMarketer reports that every GEO response differs from every other, so ongoing measurement is not a nice-to-have. It is the job.
For the marketing leader who built a career on impressions and CTR.
The investment-measurement gap
The data sources have changed. The logic has not. 94% of marketing leaders plan to increase GEO spend in 2026 (eMarketer, January 2026), yet the majority still do not invest in GEO measurement. That gap between intent and capability is the largest risk in AI brand visibility strategy right now.
Translation: everyone is about to spend money on something almost no one is measuring. This is a horror movie in polite marketing clothes.
Traditional KPIs and their GEO equivalents
Why no single number captures AI visibility
The Gist Brand Visibility Manifesto puts it simply: visibility in AI environments is presence plus attribution plus influence. High Share of Voice with low Share of Citations means you are visible but not authoritative. High Authority with low Recall means you are trusted but never recommended. The Brand Health framework captures all four dimensions in one place so marketers can diagnose exactly where the gap lives. You know what needs to be fixed, in plain English, before you spend the next quarter guessing.
The questions you were going to ask anyway.
How do you measure AI visibility?
Track the nine GEO baseline metrics Gist GEO uses to measurebrand presence, authority, and recall across AI platforms:
- Share of Voice: How often you show up in AI-generated answers.
- Sentiment: How positively AI describes you.
- Share of Citations: How often your domain is actually quoted.
- Share of Found Links: Whether AI can reach your citable pages.
- Earned Media Score: What third-party sources are saying about you.
- Citation Rate: How often your pages are part of the research set.
- Share of Recommendations: How often you are actively suggested as a best option.
- Placement: Whether you are leading the conversation or buried at the bottom.
- Average Rank in Lists: Where you land when AI ranks the field.
These feed four Brand Health dimensions: Awareness, Sentiment, Authority, and Recall.
Frequently asked questions about AI visibility
What is AI visibility?
AI visibility is the measurable presence and influence of a brand inside AI-generated answers. It measures how often ChatGPT, Claude, and Perplexity recognize, cite, and recommend your brand, with or without a click. Unlike SEO rankings, which measure blue-link position, AI visibility measures whether your brand shows up, how often, and how prominently in the answers AI systems generate.
What is Share of Voice?
Share of Voice is your share of the spotlight in AI answers. It is the percentage of AI-generated answers that mention your brand compared to competitors within a similar group of questions. It functions as the AI equivalent of traditional share of voice. Gist GEO tracks it across ChatGPT, Claude, and Perplexity as the primary input to the Awareness dimension of Brand Health.
Is AI visibility the same as SEO ranking?
Not even close. They are two different games on the same playing field. Fewer than 9% of ChatGPT and other AI citations come from URLs in Google’s top 10, per eMarketer. A brand can rank first on Google and be absent from AI answers for the same question. SEO is the foundation for crawlability and indexing. AI visibility requires separate measurement through GEO metrics.
How often should you measure AI visibility?
Every week, at the least. Gist GEO updates Brand Health scores on a weekly cadence. Only 30% of brands remain visible from one AI answer to the next (AirOps, 2025). When information is moving this quickly, your brand cannot afford to miss a moment. The Gist GEO Health Framework is designed around trends over snapshots. Every metric shows direction, delta from last period, and trend line. Always on. Always listening.
What tools exist for measuring AI visibility?
Semrush and Ahrefs offer AI visibility modules inside their SEO suites. AI-native tools like Profound and Otterly specialize in citation monitoring. Gist GEO measures all nine baseline metrics and scores Brand Health as part of a seven-dimension Health Framework. More comprehensive. Bigger impact.
The thing Goldilocks knew when she walked out of the woods.
AI is already shaping how customers perceive your brand. It is doing it right now. It is doing it while you read this sentence. It is doing it in multiple models, in every language your customers speak, in conversations your marketing team will never see because the customer never clicked anything.
You can measure it. Or you can guess at it. The marketers who learn the difference first are going to own the next era of visibility.
Rankings were the scoreboard. This is the play-by-play. And the game is already on.
All trademarks referenced are the property of their respective owners and are used for descriptive and analytical purposes only. No affiliation, sponsorship, or endorsement is implied.
