Nov 03 2026
Off

You measure your brand’s visibility in AI answers with four metrics: mention rate, citation share, position in the answer, and sentiment. Track them on a fixed set of buyer prompts each month, then read AI referral traffic in your analytics. Frequency matters most: in a SparkToro study, ChatGPT and Google’s AI repeated the same brand list less than 1% of the time.

Why does AI visibility need its own metrics?

An AI answer is generated each time, so the brands it names can change from one run to the next.

The SparkToro research, run by Rand Fishkin with Patrick O’Donnell of Gumshoe.ai and reported by Search Engine Journal in January 2026, tested 2,961 prompts.

Each prompt ran 60 to 100 times per platform on ChatGPT, Claude and Google AI Overviews. On ChatGPT and Google’s AI, the same brand list came back less than 1% of the time. The same list in the same order appeared less than 0.1% of the time.

Gumshoe.ai sells AI tracking tools, which Fishkin disclosed, and the report is not peer-reviewed. Its practical lesson still holds for B2B teams: count how often your brand appears across many runs, and read single answers as samples.

Ahrefs agrees in its guide to AI share of voice: groups of relevant prompts are usually more useful than individual answers.

Which metrics measure a brand’s visibility in AI answers?

Four metrics cover most B2B needs.

MetricWhat it measuresHow to calculate it manually
Mention rateHow often AI answers name your brandAnswers that name your brand, divided by answers checked, times 100
Citation shareHow often AI answers link to your pages as a sourceAnswers that cite your domain, divided by answers checked, times 100
Position in answerWhere your brand appears among the brands namedAverage rank of your brand in the list (1 is first)
SentimentThe tone and attributes the answer gives your brandLabel each mention positive, neutral or negative, and note the attributes used

Mention rate. Ahrefs documents its mention count this way: a brand counts once per response if it appears at least once. For example, if your brand appears in 12 of 40 answers, your mention rate is 30%.

Citation share. A citation means the answer used one of your pages as a source. Ahrefs counts citations per page and per domain, and it also records pages retrieved in the background as “Found in” a response.

Position in answer. Position shows prominence, and it moves a lot between runs. Fishkin argues that per-query ranking position in AI is unreliable, so read position as an average over many answers.

Sentiment. Sentiment shows how the answer describes you, for example specialist or reliable. Semrush documents sentiment and key narratives as part of its Brand Performance reports.

Many teams add AI share of voice. Ahrefs gives the basic formula as responses that mention your brand, divided by responses that mention any tracked brand, times 100.

How do you track AI prompts manually?

A spreadsheet and one hour a month give you a reliable baseline. Follow this checklist.

  1. Write your prompt set. List 20 to 30 questions your buyers ask, in their words. Group them by topic: category searches, comparisons, problems, and your region.
  2. Pick your engines. Choose the two or three AI tools your buyers use most, for example ChatGPT, Perplexity and Gemini.
  3. Run each prompt more than once. Use a fresh chat for each run, so earlier messages do not shape the answer. Three runs per prompt is a practical example.
  4. Log every answer. Record the date, engine, prompt, whether your brand appears, its position, cited URLs, competitors named and the tone used.
  5. Calculate the four metrics per topic. Topic groups show where you are strong and where competitors take the answer.
  6. Repeat monthly with the same prompts. Keep the wording fixed so changes come from the engines and from your content.
  7. Study the cited pages. Note which pages AI uses as sources, yours and your competitors’. Case studies with clear numbers often appear here, as we explain in why AI engines cite case studies.

When the prompt list grows, a dedicated tool often saves time.

What types of AI visibility tools exist?

AI visibility tools fall into three broad categories.

  • Prompt tracking tools run your chosen prompts on a schedule and log mentions, positions and sources. The Semrush AI Visibility Toolkit documents Prompt Tracking on platforms like ChatGPT and Google AI Mode. Its AI Visibility Score compares your mentions with the median mentions of top industry competitors, which Semrush identifies automatically.
  • Brand index tools monitor a large set of prompts and report mentions and citations against competitors. Ahrefs Brand Radar shows where your brand is mentioned in AI answers, which cited pages surface it, and your AI share of voice. Ahrefs describes counting brand mentions across 6 AI platforms.
  • Web analytics tools measure the visits AI sends to your site. Google Analytics 4 and Google Search Console cover this for most B2B websites.

Before you pick a tool, Fishkin advises asking the provider to show its math.

How do you read AI referral traffic in Google Analytics 4?

Google Analytics 4 includes an AI Assistant channel in its default channel group. Google describes it as the channel for users arriving from sources like ChatGPT, Gemini, Deepseek, Copilot and Grok. The rule is that the medium exactly matches “ai-assistant”.

Benjamin Mangold of Loves Data documents the steps. Go to Reports, then Acquisition, then Traffic acquisition. Set the dimension to Session default channel group, find the AI Assistant row, and add Session source / medium as a secondary dimension.

For a deeper view, build an Exploration with Session source in rows and add Sessions, Engagement rate, Active users and Key events. Filter Session source with “matches regex” and a pattern like the one Loves Data publishes:

^(chatgpt\.com|perplexity\.ai|gemini\.google\.com|claude\.ai|copilot\.microsoft\.com|deepseek\.com|grok\.com)$

Loves Data notes that the capitalization in the pattern must match your data. The anchors match the full source value, so check the source names in your own data.

Google’s AI Overviews and AI Mode sit outside the AI Assistant channel. Google Search Central states that they count in your overall Search Console traffic, in the Performance report under Web search.

Analytics shows part of the picture. In the Ahrefs AI traffic study, the author’s tests showed Copilot, Mistral and Jasper referrals arriving as direct traffic. Ahrefs suggests watching direct traffic on specific pages and asking “How did you hear about us?” in your forms.

The Ahrefs study (February 2025, 3,000 websites) found that 63% of sites received at least one AI visit. AI made up 0.12% of views. ChatGPT sent about 50% of that AI traffic, Perplexity about 30.7% and Gemini about 17.6%.

Adobe reports strong growth.

In its January 2026 analysis of over 1 trillion visits to U.S. retail sites, traffic from generative AI tools rose 693.4% year over year in the 2025 holiday season. AI referrals converted 31% more than other traffic sources.

This indicates a trend, it is not a forecast for your case.

Frequently asked questions

What is the best metric for AI visibility?

Mention rate is the most stable starting metric, because it counts how often your brand appears across many runs of similar prompts. SparkToro’s research suggests this frequency is more consistent than a ranking position.

How many prompts do I need to track AI visibility?

A focused set of 20 to 30 buyer prompts, grouped by topic, is a practical start for many B2B teams. Ahrefs advises reading groups of relevant prompts, because single answers vary.

Does Google Analytics 4 show traffic from ChatGPT?

Yes. GA4 has an AI Assistant channel in its default channel group for sources like ChatGPT, Gemini, Deepseek, Copilot and Grok. Some AI visits still arrive with no recorded source and appear as direct traffic.

Where do Google AI Overviews clicks appear in my reports?

Google Search Central states that they count in overall Search Console traffic, in the Performance report under Web search. In GA4, the AI Assistant channel leaves out AI Overviews and AI Mode.

How often should I measure AI visibility?

A monthly check on the same prompt set works well for many teams. Ahrefs recommends trusting trends that persist across several reporting periods over one-off spikes.

What do your AI visibility numbers tell you to do next?

Measurement starts with the four metrics from the opening: mention rate, citation share, position and sentiment, read across a fixed prompt set. Use these questions to see where you stand today.

  • Do you have a written list of the prompts your buyers ask AI?
  • Do you know your mention rate on those prompts this month?
  • Can you name the pages AI cites when it answers questions in your category?
  • Does your GA4 setup show the AI Assistant channel and its key events?
  • Do you publish proof, such as case studies with clear results, that AI can cite?

If the last answer is the weak one, proof content is often your fastest lever. Our comparison of SEO and GEO for B2B websites shows how the two work together.

casestudybook writes your case study and builds it so Google finds it and generative engines cite it. The SEO kit and AI signals are included.

Delivery takes 48 hours from the complete brief, and prices start from €99. One round of revisions is included, and you get a refund if you are not convinced. See how it works on casestudybook.com or book a short call.

Sources

Posted inGEO & AI Search