supereo_

guide // how to // 2026

What an AI visibility score is, and how to get one you can trust

An AI visibility score is one number for how often AI answers mention, cite or recommend your brand across a set of prompts. Each tool that prints one builds it its own way: Semrush scores you from 0 to 100 against others in your industry[20], while HubSpot's free grader adds up five dimensions and weights sentiment above the rest[26][27]. supereo is our pick, and we make it: it leads with how often answers recommend you, shows the range that rate likely sits in, then does the work to raise it and proves each change against control prompts. Before you trust any score, ask which prompts and engines it counts, how many answers it rests on and how wide its range is. In the real run below, the client called one of supereo's tools: get_page_diagnosis.

Updated . Prices checked 19 November 2026.

The verdict

  1. A rate with its range on it, then raised for you

    supereoour pick, and we make it

    • best for teams that want AEO done, not a dashboard: supereo does everything it takes to get you recommended, from your site and pages to the roundups, placements and videos AI reads;
    • it proves each move against control prompts, with sample sizes and ranges on every number;
    • run it from Claude or ChatGPT: an MCP server with the same tools supereo's own agent uses.

    From $149 a month

    Run the free auditFree, no card, every answer shown.

The steps

  1. Choose the prompts your buyers type, none of them naming you

    Start from what a buyer asks before they know your name: the problem, the category, the comparisons and the alternatives. Keep prompts that name you on a list of their own, because a prompt that names you gets answers about you[7]. The prompts you pick move the number more than chance does[6], so write them down and keep them fixed.

  2. Decide what the score counts

    Mentioned, recommended and cited are three counts, and a score that blends them hides which one moved. supereo's judge reads every answer three ways: it mentions you, it offers you as something to buy or use, or its sources hold a page on your site[5]. Report one as your score and keep the other two beside it.

  3. Ask every engine the same prompts, many times over

    One answer is one draw. Graphite's researchers advise against relying on a single response and suggest at least 10 runs of each prompt for a quick estimate[13]. Ask each engine the same way every time, from your buyers' country, on a fixed schedule, so every run counts alike.

  4. Read the score with its range

    Turn the counts into a rate and put a range around it: the band the true rate likely sits in, given how many answers you have. supereo's range is 1.96 standard errors either side of the rate[2], and a rate whose range is still too wide shows the range and no number at all[3].

  5. Compare like with like

    Add prompts, drop an engine or change the competitors in the set, and the number moves for reasons that have nothing to do with you. supereo moves a number up or down only like for like[8], so a change in what you track never reads as a change in how AI answers.

  6. Credit a change only against control prompts

    Before you say a page raised your score, look at prompts it could not have touched. supereo measures a change as how much its target prompts moved minus how much your other prompts moved[9]; the other prompts carry the model updates, competitors' launches and slow weeks that move everything at once[10].

Real runs

a real run // 19 November 2026

One page scored for one question by supereo's page diagnosis

Recorded from supereo's MCP server on its staging environment, https://staging.supereo.app/api/mcp. 1 call.

get_page_diagnosis · 2,776 ms ·

Sent

{
  "url": "https://supereo.app/guides/how-to-measure-ai-visibility",
  "query": "ai visibility score"
}

Received

{
  "url": "https://supereo.app/guides/how-to-measure-ai-visibility",
  "finalUrl": "https://supereo.app/guides/how-to-measure-ai-visibility",
  "status": 200,
  "problem": null,
  "kind": {
    "value": "other",
    "source": "content_read"
  },
  "pageType": "other",
  "score": 100,
  "checks": [
    {
      "id": "server_rendered",
      "label": "Real text in the HTML, not a JavaScript shell",
      "pass": true,
      "detail": "3923 words in the HTML"
    },
    {
      "id": "fresh",
      "label": "Updated within 30 days",
      "pass": true,
      "detail": "newest date 2026-11-18 (json_ld), 1 day old"
    },
    {
      "id": "specific",
      "label": "Specific rather than generic",
      "pass": true,
      "detail": "specificity 4 of 5, generic AI style 3 of 5"
    }
  ],
  "gaps": [],
  "numbers": {
    "dollarFigures": 3,
    "percentFigures": 11,
    "newestDate": {
      "date": "2026-11-18T00:00:00.000Z",
      "source": "json_ld",
      "ageDays": 1
    }
  },
  "read": {
    "status": "done",
    "answersUpFront": "partly",
    "specificity": 4,
    "originalData": true,
    "sourcesCited": true
  },
  "costCents": 0.0743
}

Trimmed: the content read's summary, a model's one-line description of the page: its checks are above.

get_page_diagnosis read one of supereo's own guides as a crawler without JavaScript gets it and scored it on the checks the pages AI engines cite tend to pass. The score is about the page, whether AI can read it and lift an answer from it, and says nothing of how often an AI answer names the brand.

What each tool's AI visibility score counts

Most scores start from one count: run a set of prompts, read the answers, and see how often your brand turns up. Graphite, whose researchers proposed running each prompt many times, calls that count visibility[12]. The tools differ in what they divide by and what they add on top.

Peec AI's visibility counts the AI responses where your brand appears, out of all its tracked runs[32]. Profound's Visibility Score looks the same but divides by a smaller number: only the responses that mention at least one brand[33], so an answer that names no brand at all doesn't pull it down.

HubSpot's AEO tool reports brand visibility as how often you appear in the responses to your tracked prompts[34], beside a prompt coverage figure that counts prompts instead of answers: the prompts where you appear at least once[35]. Its free grader is a different instrument again, scored out of 100 from five dimensions with sentiment carrying the most weight[26][27].

Semrush's score runs from 0 to 100 and combines two factors: how many topics include your brand, and how consistently you are mentioned within them[37]; its paid toolkit also benchmarks you against your competitors' median and refreshes the score daily[21]. Rankscale's Visibility Score combines how often you are found with how high you rank when found[38], so the same mentions score lower when they come late in an answer.

Writesonic's AI Visibility Scores count how often you appear in AI answers, split into branded and non-branded prompts[29], and Profound averages its score over prompts it runs every day[28]. Ahrefs calls its numbers directional, modeled signals rather than exact counts[30]. supereo counts recommendations, not mentions, and puts a range on every rate, as the steps above show.

Why two tools give the same brand two different scores

Point two tools at one brand and you will usually get two numbers, and neither has to be wrong. They run different prompts, and the choice of prompts moves a number more than run-to-run luck does[6].

They ask the engines in different ways. Graphite found that answers from an engine's API, from its logged-out app and from a logged-in account can vary a lot[19], and a grader that reads what a model learned in training measures something else again[25].

They count different things: a mention, a recommendation, a citation, a place in a list, a sentiment, and even two counts of mentions can divide by different totals, as Peec AI's and Profound's do[32][33]. Some scores are relative, too: a score benchmarked against your competitors' median[21] can fall when a competitor gains, with your own answers unchanged.

So compare a tool's score only with the same tool's earlier score, on the same prompts and engines. Two tools' numbers side by side tell you about the tools as much as about you. How AI visibility trackers sample lays out each tracker's choices.

Why a free one-time score moves from day to day

A one-time check asks each of its prompts once, and AI answers rarely hold still between runs. Graphite showed how far one response can mislead: in a single answer, a flavor that turned up in only a handful of 200 responses looked as common as one that turned up in every response[14].

Some free scores don't ask the engines live at all. HubSpot's grader reports what ChatGPT, Perplexity and Gemini say about you from their training data[25]: a reading of what the models learned, not of what their live search turns up this week.

supereo's own free audit is a one-time check as well: it runs its 20 prompts once each on five engines[11]. It shows you where you stand and every answer behind that, not a rate you can watch move.

So read any free score as a snapshot of one day. Run the same prompts again on a schedule, many times each, before you call a move a trend.

How wide a score's range is, and when a change is real

A score built from a sample has a margin around it. In Graphite's worked example, a flavor seen once in the first ten answers could, by its confidence range, be anything from rare to fairly common[15].

The range narrows as answers pile up, and it is widest for a brand that turns up in about as many answers as it misses[17]. Graphite's figures: 10 responses only separate rates that are far apart, and 50 responses still can't separate rates that sit close together[16].

That is why supereo puts a badge on every rate: Solid when the range is within 5 answers in 100 either side, Rough within 10, and Early when it is wider, which shows the range and no number[3].

A change is real when the ranges before and after it don't overlap[18], and when it beats what your control prompts did over the same weeks. A score that moves by less than its range has, as far as anyone can tell, moved by luck.

A page score is not an AI visibility score

The real run above asked supereo's page diagnosis to score one of supereo's own guides for the query ai visibility score. The page passed every check it ran: its words are in the HTML a crawler gets, it was updated recently, and it reads as specific rather than generic. The diagnosis still judged the page's answer to be only partly up front.

That is a page score. It says whether an AI engine that reads the page can lift an answer from it. It says nothing about how often AI answers name or recommend supereo when someone asks about AI visibility scores: only running those prompts and counting the answers can tell that.

Several tools sell a score of this kind beside their visibility numbers: Otterly's GEO score rates one URL's readiness for AI search[31], and Semrush's AI Search Health scores how optimized a whole site is for it[24]. Use a page score to find what to fix on a page, and a visibility rate to see whether fixing it changed the answers.

An AI visibility score is not an SEO visibility score

Rank trackers have printed a visibility score for years, and it measures something else: how well your pages rank in classic search results. An AI visibility score reads the answers instead. Semrush keeps the two apart, with a switch between AI Visibility and the traditional SERP-based Regular Visibility[22].

The two can move apart. A page can rank well and still not be the source an AI engine quotes, and a brand can be named in answers for questions where none of its pages rank. So a strong SEO score beside a weak AI one can mean the engines find your pages but take their answer from someone else's, which is a question of what your pages say and who else writes about you, not of indexing.

Read them side by side and never add them together: one tells you whether people can find your pages, the other whether AI answers send buyers your way.

Have it done: supereo

01supereoour pick, and we make it

  • best for teams that want AEO done, not a dashboard: supereo does everything it takes to get you recommended, from your site and pages to the roundups, placements and videos AI reads;
  • it proves each move against control prompts, with sample sizes and ranges on every number;
  • run it from Claude or ChatGPT: an MCP server with the same tools supereo's own agent uses.

From $149 a month

  • Does the worksupereo writes those pages for you, aimed at the questions where AI skips you. supereo.app
  • Proves it against controlsEach change has a few target questions, and we compare them with your other questions, the ones no change targets. supereo.app
  • Runs from Claude or ChatGPTRun the same loop from Claude, Cursor or Codex. Your agent gets every tool supereo's own agent uses: your buyers' questions, every answer, the pages AI cites, your brand facts. supereo.app
  • Free auditfree audit // about 2 minutes // no card supereo.app
PlanA monthPromptsAI engines
Autopilot$14940ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Claude

Prices from supereo.app, checked 19 November 2026. What it does, from its own pages, checked 16 November 2026.

How the proof works, with every number it runs on: How supereo measures.

Questions

What is an AI visibility score?

A number for how often AI answers mention, cite or recommend your brand across a fixed set of prompts. Each tool builds its own, so the useful questions are which prompts and engines it counts, what counts as a hit and how wide its range is. supereo leads with how often answers recommend you, with a range on every rate.

How is an AI visibility score calculated?

Usually as the answers that include your brand divided by all the answers counted, across a set of prompts and engines, which is Graphite's definition of visibility[12]. Some tools add more on top: Semrush compares you with others in your industry[20], HubSpot's grader sums five weighted dimensions[26], and Rankscale also weighs how high you appear in each answer[38].

What is a good AI visibility score?

There is no bar that holds across tools, because each one counts differently. HubSpot, for one, calls a score above 67 high and below 34 low on its own scale, and says the score works best as a baseline for tracking improvement[36]. Compare yours with the competitors named in the same answers, on the same prompts, and watch whether it moves past its range over a few weeks.

Can I check my AI visibility score for free?

Yes: Semrush's free checker needs no account and allows three runs a day[23], and Ahrefs offers one with no signup[40]. Each is a single reading that the next run can move, and our free AI visibility checkers guide compares them. supereo's free audit shows every answer behind its reading, judged for whether it mentions, recommends and cites you: run the free audit.

Is there an app that shows my AI visibility score?

Yes, and you may not need a separate one: supereo's MCP server carries the same tools its own agent uses, so Claude or ChatGPT can run it in a chat. To see where you stand today, with every answer, run the free audit.

Why did my AI visibility score drop?

First check whether it dropped past its range: a move inside the range is luck as far as anyone can tell. Then check what changed in the count itself, such as new prompts, a different set of competitors, or an engine added or removed. Only then read the answers, for a competitor that took your place or an engine that stopped citing your pages.

How do I improve my AI visibility score?

Open your site to the AI crawlers, put the answer to each buyer question at the top of your own pages, earn a place on the lists, reviews and comparisons the engines quote, and state your facts the same way everywhere. Each engine has its own route: how to get recommended by Claude and how to get cited by ChatGPT. supereo does that work for you and proves each change against control prompts.

How many answers does a reliable AI visibility score need?

Enough per prompt that the range is narrower than the change you want to see. Graphite's starting point is at least 10 runs of each prompt[13], and more for a brand that turns up in about as many answers as it misses[17]. How to measure AI visibility walks through the rest.

Does an AI visibility score include Google AI Overviews?

Only if the tool asks Google. supereo counts AI Overviews and AI Mode among its six engines. For Google alone, Search Console reports how often your pages appeared in its AI features and which pages[39], though not what each overview said.

Is a GEO score the same as an AI visibility score?

No. A GEO score like Otterly's rates one page's readiness for AI search[31], while a visibility score counts what the engines' answers say about your brand. The real run on this page shows the difference.

How we compiled this

Where this page's words say how an AI engine finds, reads or cites pages, the sentence links the engine's own documentation, or another source, and the day it was read.

The runs are real calls to supereo's MCP server on its staging environment, made as a house account, recorded as they happened: the arguments as sent, each answer as the client received it, its JSON indented and trimmed only where it says so, the time each call took and when it ran.

We make supereo, so it is the one tool this page names as a pick.

How supereo measures its own results, with every number it runs on: How supereo measures.

supereo (supereo.app) is an AI agent that does everything it takes to get your brand recommended by ChatGPT, Gemini, Perplexity, Claude and Google, on your site and off it: site fixes, new pages, roundups, placements and video. It proves what worked against control prompts. Made by superagnt.

Sources

  1. A rate is the share of the last 4 weeks of answers that recommend you, shown as answers out of every 100. supereo.app, read 19 November 2026.
  2. Its range is where it likely sits, its 95% range: 1.96 standard errors either side. supereo.app, read 19 November 2026.
  3. Solid: the range is within 5 answers in 100 either side. Rough: within 10. Early: wider than that. An Early rate shows its range and no number. supereo.app, read 19 November 2026.
  4. Named: the answer mentions you. Recommended: the answer offers you as something to buy or use, with the answer's own words kept as its quote. Linked: the answer's sources hold a page on your site, matched by its address. supereo.app, read 19 November 2026.
  5. Which prompt is run moves a number more than chance does supereo.app, read 19 November 2026.
  6. The two sets are counted apart and never added together, since a prompt that names you gets answers about you. supereo.app, read 19 November 2026.
  7. A number moves up or down only like for like supereo.app, read 19 November 2026.
  8. Added by the change: how much the target prompts moved, minus how much your other prompts moved, on each AI against its own before, then averaged over the AIs, each counting the same. supereo.app, read 19 November 2026.
  9. They show how AI answers moved on their own: a model update, a competitor's launch, a slow week. supereo.app, read 19 November 2026.
  10. The free audit runs its 20 prompts once each, on five of these engines: every one but Claude. supereo.app, read 19 November 2026.
  11. This is now commonly referred to as visibility: the percentage of responses in which a brand appears. graphite.io, read 19 November 2026.
  12. Do not rely on a single response. Instead, run each prompt at least 10 times for a quick estimate. graphite.io, read 19 November 2026.
  13. If we look only at one response (as some tools do), for example, the one in the rightmost column of the table above, we would think that s’mores appears as often as vanilla. In reality, s’mores appears in only 8 of the 200 responses, while vanilla appears in every response. graphite.io, read 19 November 2026.
  14. With an estimate of 10%, we can be 95% confident that the true visibility for s’mores is between 1.79% and 40.42%, and that the true visibility of vanilla is between 72.25% and 100.00%. graphite.io, read 19 November 2026.
  15. For example, with 10 responses, we can statistically distinguish between 75% and 25%, but not 75% and 50%. With 50 responses, we can statistically distinguish between 50% and 30%, but not 50% and 40%. graphite.io, read 19 November 2026.
  16. Visibility confidence intervals are wider when the visibility is close to 50%. The reason is that, statistically, visibility values near 50% require more responses to estimate, whereas values close to 0% or 100% are easier to estimate. graphite.io, read 19 November 2026.
  17. If the confidence intervals do not overlap, the difference is statistically significant. graphite.io, read 19 November 2026.
  18. Responses from APIs, logged-out accounts, and logged-in accounts can vary significantly. graphite.io, read 19 November 2026.
  19. AI Visibility Score: A 0 to 100 metric that represents how often your brand appears in AI-generated answers compared to others in your industry. semrush.com, read 19 November 2026.
  20. AI Visibility Score is Semrush's proprietary metric that measures how often your brand appears in AI answers, automatically benchmarked against your competitor median. It updates daily and covers worldwide and regional breakdowns. semrush.com, read 19 November 2026.
  21. You can toggle between AI Visibility and Regular Visibility (traditional SERP-based) using the visibility selector at the top of the interface. semrush.com, read 19 November 2026.
  22. Yes, the AI Visibility Checker is free to use. You can use the tool three times per day for free, without registering or creating an account. semrush.com, read 19 November 2026.
  23. AI Search Health is a score that reflects how optimized your website is for AI search. semrush.com, read 19 November 2026.
  24. HubSpot's AI Search Grader is a free, one-time check that reveals what ChatGPT, Perplexity, and Gemini say about you based on their training data. hubspot.com, read 19 November 2026.
  25. HubSpot's AI Search Grader evaluates your brand across five dimensions, each contributing to your overall score out of 100. hubspot.com, read 19 November 2026.
  26. Sentiment carries the most weight because it measures how AI characterizes your brand, not just whether it recognizes you. hubspot.com, read 19 November 2026.
  27. AI platforms don't return the same answer twice. We run every tracked prompt daily so your visibility score reflects your true average across all responses. tryprofound.com, read 19 November 2026.
  28. AI Visibility Scores How often you appear in AI answers. Daily tracking. All platforms. Broken down by branded vs. non-branded, commercial vs. informational. writesonic.com, read 19 November 2026.
  29. Metrics are directional indicators, not exact traffic counts – best understood as modeled visibility signals, and not performance metrics. ahrefs.com, read 19 November 2026.
  30. Check any URL for AI-search readiness. Get a structured GEO score and actionable optimization tips. otterly.ai, read 19 November 2026.
  31. Visibility — "Percentage of AI responses where your brand appears." The share of tracked prompt runs in which the brand shows up at all. peec.ai, read 19 November 2026.
  32. The percentage of responses that mention your brand, out of all responses that mention at least one brand. docs.tryprofound.com, read 19 November 2026.
  33. Brand visibility measures how frequently your brand appears in AI-generated responses for your tracked prompts. hubspot.com, read 19 November 2026.
  34. Prompt coverage: the percentage of prompts where your brand appears at least once. knowledge.hubspot.com, read 19 November 2026.
  35. Scores above 67 are considered high, 34 to 66 medium, and below 34 low. Most brands start in the low to medium range. The score is most useful as a baseline to track improvement over time rather than a fixed benchmark hubspot.com, read 19 November 2026.
  36. AI Visibility measures how often and how prominently a brand appears in AI-generated answers. The score (0–100) combines two factors: Topic Coverage — how many topics include your brand in AI answers, compared to all other domains. Mention Consistency — within those topics, how frequently your brand is mentioned across all responses. semrush.com, read 19 November 2026.
  37. It combines two signals: how often your brand is found (Detection Rate) and how high it ranks when found (Average Position). help.rankscale.ai, read 19 November 2026.
  38. Impressions: How often URLs from your site appeared in generative AI features in Search and Discover. Pages: Check which URLs appeared within AI features. developers.google.com, read 19 November 2026.
  39. Free AI Visibility Checker See how your brand shows up in AI search results from ChatGPT, Gemini, Perplexity, Copilot, and Google AI Overviews. No signup required. Results in seconds. ahrefs.com, read 19 November 2026.

install with one prompt // any AI agent

Install supereo with one prompt.

Paste one line into the agent you already use. Coding agents set supereo up themselves, and Claude and ChatGPT walk you through adding it. You sign in once, then ask it how AI sees your site.

setup prompt

Install supereo for me. It's the supereo MCP server at https://mcp.supereo.app/mcp (on Claude it's a connector I add in my browser; elsewhere I sign in through my browser). The setup guide for each kind of agent is at https://supereo.app/install
  1. 01copy the prompt
  2. 02paste it into your agent
  3. 03sign in once, in your browser

or wire it by hand

https://mcp.supereo.app/mcp
claude · ~/dev/acme
Claude Code v2.1.206

Welcome back Alex!

Opus 5.5 · Claude Max · ~/dev/acme

⏸ manual mode on · ? for shortcuts

An example session. The same prompt works in every client.