AI discovery monitoring and content

How to measure brand visibility in AI answers

8 min read


On this page

The short answer

Measure brand visibility in AI answers as a rate, not a snapshot. Run a fixed set of buyer questions many times, count how often each engine names you, and set that against your rivals. Check that the answer describes you correctly, and read Search Console and Bing's free AI Performance report for the clicks and citations they record.

Key takeaways

  • AI tools gave the same list of brands twice under 1 in 100 times in SparkToro's study, so a single check proves little.
  • Visibility rate is the share of tested prompts that name your brand, and citation share sets it against rivals, as HubSpot defines them.
  • Your place in an AI answer's list moves from run to run, so a rate over many runs tells you more than an order.
  • Google counts AI Overviews and AI Mode inside Search Console's Web search type, and Bing's AI Performance report shows Copilot citations at no cost.
  • Clicks undercount AI visibility: Pew found people clicked a result link on 8% of visits when an AI summary appeared.

What metrics matter most for brand visibility in AI answer engines?

Four metrics hold up: how often you are named, how that compares with rivals, whether the answer is correct, and how it frames you. HubSpot's guide calls the first the AI visibility rate, which "measures how often your brand appears in AI-generated answers across a defined set of prompts", and the second citation share, your brand's percentage of citations relative to competitors across the same prompt set. It treats accuracy and sentiment as qualitative checks: is the engine describing your product, pricing or use cases correctly, and is the framing positive, neutral or negative. The research behind this points the same way. After thousands of prompt runs, SparkToro's Rand Fishkin concluded that visibility percentage across dozens to hundreds of prompts, run multiple times, is a reasonable metric. A rate on its own is half the picture. Being named often with the wrong price is worse than being named less often and described correctly, which is why the accuracy check sits next to the count.

How do you work out an AI visibility rate?

Divide the prompts where your brand appears by the prompts you tested, and multiply by 100. That is the formula HubSpot gives, and the same idea runs through SparkToro's results. When Google's AI was asked to recommend digital marketing consultants with expertise in e-commerce, one agency appeared in 85 out of 95 responses. In ChatGPT's answers on West Coast cancer care, one hospital showed up in 69 of 71 answers, which the study reports as a 97% visibility rate. A rate only compares fairly inside one market. In the same research, the top headphone brands often reached 90 to 100% visibility, while a prompt asking for brand design agencies produced top rates in the 30s and 40s. A wide field with many possible answers spreads the mentions thinner. So the same rate can be strong in one trade and weak in another, and the useful comparison is your own rate against the rivals named for the same questions.

Why does one check of ChatGPT tell you so little?

Because the same question rarely returns the same brands twice. In SparkToro's study, 600 volunteers ran 12 prompts through ChatGPT, Claude and Google's AI a combined 2,961 times. The study found under a 1 in 100 chance that ChatGPT or Google's AI gives the same list of brands in any two responses, and about 1 in 1,000 runs before two lists come back in the same order. The number of brands on each list changed too, from as few as two or three to ten or more. A screenshot of one answer, good or bad, is one draw from that spread. Fishkin's advice for anyone who wants to know an AI's real set of recommendations is to ask over and over, usually at least 60 to 100 times, and average the results. Few owners will do that by hand. The practical lesson is to read any single answer as a sample and to judge visibility on the pattern over weeks, not on one bad day.

Does your position inside an AI answer mean anything?

Very little. The same study found the order of brands so unstable that Fishkin wrote that any tool giving a "ranking position in AI" is full of baloney. The cancer care example shows why: the hospital named in 69 of 71 ChatGPT answers was the top mention in only 25 of them. Being in the set is stable; being first is not. Google's own reporting makes a related point. In Search Console, an AI Overview occupies a single position in search results, and all links in the AI Overview are assigned that same position. So the position Search Console shows for a link inside an AI Overview tells you where the overview sat on the page, not where your link sat inside it. Track whether you are named and how often, and leave the order aside.

Which prompts should you measure, and whose words should they use?

Measure the questions your buyers ask, in many wordings, and run the same set for your competitors. HubSpot describes a prompt set as a curated list of questions that reflect how your target audience actually searches in AI engines, and advises running it for competitors as well as your own brand. Real wording varies far more than search keywords. When SparkToro asked people to write their own prompt for one buying need, the 142 answers had an average semantic similarity of 0.081. Even so, across 994 responses to those prompts, the same leading headphone brands showed up 55 to 77% of the time. The engines read the intent beneath the words. Full questions also matter on Google. Pew Research found that 60% of searches beginning with words such as "who", "what", "when" or "why" produced an AI summary, against 8% of one- or two-word searches. Write your prompts the way a buyer asks a question, not the way an SEO tool lists a keyword.

Can Google Search Console show your visibility in AI Overviews and AI Mode?

Yes, as part of your normal search figures. Google's documentation says sites appearing in AI features such as AI Overviews and AI Mode are included in the overall search traffic in Search Console, reported in the Performance report within the "Web" search type. They sit in the same totals as your other Google results. The Search Console help page sets out the counting rules: clicking a link to an external page in an AI Overview or in AI Mode counts as a click, standard impression rules apply, and a follow-up question in AI Mode counts as a new query. Eligibility is the same as for any search result. A page must be indexed and eligible to be shown in Google Search with a snippet, and Google says there are no additional technical requirements. For an owner, this makes Search Console the free baseline. If a page is not indexed, it cannot be a supporting link in an AI Overview at all, so the Pages report is the first place to look.

Where can you see which of your pages AI answers cite, for free?

Bing Webmaster Tools has the clearest free view, for Microsoft's AI answers. Its AI Performance report, introduced in February 2026, shows how your content appears across Microsoft Copilot, AI-generated summaries in Bing and select partner integrations. It reports total citations, the average number of your pages cited per day, the grounding queries the AI used to find your content, and citation counts for each URL. In June 2026 Bing added Citation Share, calculated as the percentage of citations attributed to your site out of all citations shown across all sites for that same grounding query. Bing is careful about its limits: the metric does not expose competitor domains, represent traffic share, or assign quality scores to content. It covers Microsoft's own AI answers and select partners, so it is one reading among several. Its grounding queries are useful on their own, since they show the questions for which an AI system already reached for your pages.

Why measure mentions when AI answers send so few clicks?

Because the answer itself is where many buyers stop. Pew Research found that Google users who saw an AI summary clicked a traditional result link in 8% of visits, against 15% for those who did not. They clicked a link inside the summary itself in just 1% of visits, and they ended their browsing session on 26% of pages with an AI summary, against 16% without. If you judge AI visibility by referral clicks alone, you miss most of the moments when a buyer read your name, or a rival's. Google adds one point in favour of the clicks that do come: it says that people who click from results pages with AI Overviews are more likely to spend more time on the site. Count the mentions to see whether you are in the answer, and count the clicks to see what that presence is worth.

How does ReachKit measure brand visibility in AI answers?

ReachKit checks the answers every week and then writes the pages meant to change them. Its home page says: "Every Monday we ask Google’s AI and ChatGPT your buyers’ questions. You see if they name you, or a competitor." Under Google and ChatGPT tracking it lists "Where your pages sit in Google, and which ones Google’s AI and ChatGPT link to.", with up to five competitors "Watched every week." The how it works page puts the research step plainly: "We look up what your buyers search and ask AI, and see who gets chosen instead of you." The measuring is there to choose the writing: the pricing page lists "Up to 7 pages a week, new or updated. Never filler." for €49 a month, tax included. It also lists Perplexity under what ReachKit doesn't do, so that engine is not part of the check. Read against SparkToro's findings, a weekly reading over the same buyer questions is a trend to follow over months, not a verdict on any one Monday.

Run the free scan to see which buyer questions name a rival in AI answers and not you.

Sources