Updated on 12 November 2026, by João Leite, Founder of murmur.

The AI visibility score is the number the tool gives the brand in the dashboard, with a different calculation in each tool. In Semrush, it is a score from 0 to 100 that compares the brand’s mentions in AI answers with the median of the sector’s main competitors. In HubSpot, it is the share of answers that cite the brand: cited in 7 of 10 answers, the score is 70%. That is why two dashboards give different numbers for the same brand, and the score is only comparable with itself, on the same questions, over time.

  • Each tool does its own calculation. The AI visibility score does not have a single definition, and the same name hides different calculations.
  • The Semrush score goes from 0 to 100 and compares with competitors. It looks at how many times the brand is mentioned against the median of the sector’s main competitors, and it is not a percentage of answers.
  • The HubSpot score is the share of answers. On the HubSpot AEO page, a brand cited in 7 of 10 answers has a score of 70%.
  • In Bing Webmaster Tools, the calculation is of citations of the site, with no score. Microsoft writes that Citation Share is an observation metric, not a ranking, and that it does not give a quality score.
  • No number is exact, and the way out is to repeat. Semrush writes that no platform gives the exact number; in Otterly, the variation of the measured number fell by about 73% when the list went from 10 to 100 questions.

The AI visibility score is a different calculation in each tool

The AI visibility score is a different calculation in each tool, and the same name does not mean the same number. Every score starts from the answers the AI models gave to a set of questions, and the difference lies in three choices: what goes into the calculation (the brand name, the site link or both), what the number is compared against (the total number of answers or the competitors) and on what scale it comes out (percentage or score). Two tools that make different choices give different numbers for the same brand, in the same week. Three tools make these choices visible on their own page, each in its own way: Semrush, HubSpot and Microsoft’s Bing Webmaster Tools.

The simplest calculation is presence. Presence is how many answers the brand appears in, divided by the total number of answers. The market also calls this share of voice, and the difference between presence and share of mentions is in the guide on what share of voice in AI is.

The Semrush score from 0 to 100 compares the brand with competitors

The Semrush score from 0 to 100 compares the brand’s mentions with those of the sector’s competitors, and it is not the percentage of answers in which the brand appears. In the help section for the AI visibility toolkit, Semrush writes that “The AI Visibility Score reflects how often your brand is mentioned in AI-generated answers compared to the median number of mentions for your top industry competitors” (Semrush AI visibility toolkit help). In the free checker in Portuguese, the same score appears as a metric from 0 to 100 that shows how often the brand appears in AI answers compared with others in the sector (Semrush free checker).

The Semrush score, then, depends on who the competitors are. The same brand, with the same mentions, goes up if the sector’s competitors appear less and goes down if they appear more. The score says how the brand stands against the median of the main competitors, not in how many answers it appears.

In HubSpot, the score is the share of answers that cite the brand

In HubSpot, the visibility score is the share of answers that cite the brand, in percentage. The HubSpot AEO page gives the example: “se a sua marca é citada em 7 de 10 respostas, sua pontuação de visibilidade é 70%” (“if your brand is cited in 7 of 10 answers, your visibility score is 70%”) (HubSpot AEO page). HubSpot, which sells CRM, divides by the total number of answers, and in the page’s example no competitor enters the calculation.

This calculation is close to presence, because it divides the answers with the brand by the total number of answers. The weight of each answer depends on which questions entered the list and how many there were, and that is why the same brand can have 70% in one list and another number in another.

In Bing Webmaster Tools, the citation metric is neither a score nor a ranking

In Bing Webmaster Tools, the calculation starts from the site cited as a source, and it does not come out as a score. On 16 June 2026, Microsoft put Citation Share into preview, inside the report that shows where the site’s content is cited in the AI answers of Copilot, Bing and selected partners, and writes that the metric does not show competitors’ domains, does not represent a share of traffic and does not give a quality score (Microsoft’s announcement on the Bing blog). For Microsoft, Citation Share is an observation metric, not a ranking. The rest of the report is in the guide on what Search Console and Bing Webmaster Tools show about AI performance.

Why two dashboards give different scores for the same brand

Two dashboards give different scores for the same brand because each one does its own calculation, with its own questions. Semrush compares the brand with the median of competitors, on a score from 0 to 100; HubSpot divides the answers that cite the brand by the total number of answers; Bing Webmaster Tools looks at where the site was cited, with no competitor and no score. Presence is how many answers the brand appears in, divided by the total number of answers.

In the table, presence, the calculation murmur uses, comes first, with the same fields as the three tools.

Presence, murmur’s calculationSemrushHubSpotBing Webmaster Tools
Scalepercentagescore from 0 to 100percentageshare of citations, with no quality score
What goes into the calculationthe answers in which the brand appearshow many times the brand is mentioned in AI answersthe answers that cite the brandwhere the site’s content is cited in AI answers
Against whatthe total number of answersthe median of mentions of the sector’s main competitorsthe total number of answersdoes not show competitors’ domains

The number also changes depending on whether the tool counts the brand name in the answer or the site link as a source, and the difference between mention and citation is in the guide on Semrush and Ahrefs measure citation in AI. Before comparing two scores, check four things:

  1. The scale: score from 0 to 100 or percentage of answers.
  2. What goes into the calculation: the brand name, the site link or both.
  3. What the number is compared against: the total number of answers or the competitors.
  4. Which questions, how many and in which AI models the tool ran.

A score that passes the four checks is comparable with itself over time. The comparison of Semrush with a dedicated tool is in the guide Peec AI or Semrush AI Visibility, and what to ask before trusting any number is in the guide on the reliable metric and how to check any number.

No number is exact, and that is why the score is compared on the same questions over time

No score is an exact number, and that is why it is only comparable on the same questions, rerun over time. Semrush writes, in the help section on the origin of AI visibility data, that AI answers change fast and are highly personalized, and that this is why no platform gives exact visibility numbers. The answer to that is to always measure the same list of questions, each one in a new conversation, and to look at the direction of the number from one measurement to the next.

The list also cannot be short. Otterly, which sells AI monitoring, calls the number it measures Brand Coverage, and in a set of 320 questions, in a study updated in September 2026, the variation of that number fell by about 73% when the list grew from 10 to 100 questions. How many questions to use is in the guide on how many questions to use to measure brand presence, and how often to measure is in the guide on how often to measure brand presence in AI models.

At murmur, every number comes with its calculation

At murmur, a GEO agency in Brazil that places brands in the answers of ChatGPT, Claude, Google AIO and other AI agents, every number comes with its calculation. The usual calculation is presence. Presence is how many answers the brand appears in, divided by the total number of answers. Every week, murmur redoes the same list of questions in ChatGPT, Claude, Perplexity and Google AIO. The dashboard puts the brand’s presence next to that of competitors and is included in the service, not sold separately; storing the answer to each question is one of its functions.

The case study of Mindo, a B2B motion design studio in São Paulo, shows the calculation behind each number. On 25 and 26 June 2026, ChatGPT recommended Mindo in 12% of the questions in its market; on 13 and 14 July, in 23%. On the same questions and dates, first place in the answer went to Mindo in 16% of them, against 3% in June.

The number serves the work, and the person who reads these numbers is the same person from start to finish: they study the company and the sector, see what the AI models answer and which brands appear in its place, take care of the content that changes the answer and follow the result. murmur’s measurement runs on a dedicated server in Brazil, always from the same internet address and with the browser in Portuguese, not on a team member’s computer. That way the city of whoever is measuring does not affect the answer, and one week can be compared with the next.

Frequently asked questions

Is a score of 70 good?

A score of 70 only says something with the calculation and the sample next to it. In Semrush, 70 is a score from 0 to 100 that compares the brand’s mentions with the median of the sector’s main competitors; in HubSpot, 70% means the brand was cited in 7 out of 10 answers. The number gains meaning when you know which questions, how many and in which AI models went into it, and when it is compared with the same list on another date and with competitors on the same questions.

Can you compare the Semrush score with the HubSpot one?

The Semrush score and the HubSpot one cannot be compared directly, because they make different calculations. The Semrush one is a score from 0 to 100 against the median of the sector’s competitors, and the HubSpot one is the percentage of answers that cite the brand. Each is compared with itself, on the same questions, over time.

Does the score go up if the AI only cites the site, without naming the brand?

Each tool answers in its own way. In Semrush, the score counts how many times the brand is mentioned in the answers, according to the help page of the AI visibility kit. In Bing Webmaster Tools, Citation Share starts from where the site’s content is cited in AI answers, and there the site as a source is what counts. In HubSpot, the example on the page is that of the brand cited in 7 out of 10 answers. The difference between mention and citation is in the guide on whether Semrush and Ahrefs measure citation in AI.

How often does the score change?

The score can change with every measurement, because AI answers change fast, as Semrush itself writes. What matters is the direction of the number when the same list of questions is repeated, and the pace is in the guide on how often to measure brand presence in AI models. murmur repeats the same questions every week, in the four AI models it monitors.

Where to start

Start by noting down, for the score you already follow, the scale, what goes into the calculation, what it is compared against and which questions the tool runs. The first number without a contract comes from murmur’s free diagnostic: with the site address, no sign-up, it shows in less than two minutes how the AI models answer the questions in the company’s market, and the full report follows by email. To see the calculation applied to an entire market, murmur’s sector studies bring the brands that the AI models recommended in each sector measured, with the sample and the date of the run.

Read also

References