Mention Rate
Mention rate is the share of a fixed set of test questions where an AI assistant names your brand at all, expressed as a percentage. Ask ChatGPT, Claude or Gemini 50 realistic questions from your category and your name shows up in 9 of them, your mention rate is 18 percent. It is the most basic AI-visibility metric there is: before an assistant can recommend you or cite you as a source, it has to say your name.
Why the mention rate matters
Classic search gave you ten blue links and a chance at position two. A generative answer gives the user one paragraph. If your name is not in it, you are not part of the conversation, full stop. That shift is not a rounding error: Google's own AI Overviews now reach more than 2 billion monthly users across 200-plus countries, ChatGPT has roughly 900 million weekly users, and Gemini has passed 1 billion monthly users. At that scale, being absent from the answer is not a minor loss of a few clicks — and the click you do lose is a real one, since Pew Research found people click through to a traditional result about half as often when an AI summary is present. Mention rate turns "are we even in the conversation" into a number you can track over time, per assistant, instead of a feeling.
How it works
You start with a fixed list of realistic, need-based questions from your field — not "what does Company X do" but "which tax advisors in Cologne specialize in startups". Because AI answers vary run to run, you ask each question several times and count how many of the answers name your brand at all; 100 questions with 18 mentions gives you an 18 percent mention rate. Run a clean baseline before you change anything, so later measurements mean something. Measure each assistant separately and keep them separate in your reporting: ChatGPT, Claude, Gemini and Perplexity draw on different sources and different retrieval logic, so your rate on one can look nothing like your rate on another. An Ahrefs analysis of roughly 75,000 brands found that how often a brand is mentioned across the web correlates with AI citation rate at about 0.664 — nearly three times the correlation seen for backlinks, at about 0.218. That is the practical reason mention rate is worth tracking on its own: getting talked about, not just linked to, is the stronger lever.
Common mistakes
The most common mistake is measuring once. A single run is noise, not a number, because assistants answer the same question differently each time; ask several times and average. Second, people conflate mention rate with tone — being named is not the same as being praised, and only sentiment analysis tells you which one you got. Third, many test only branded questions instead of neutral, need-based ones; what actually matters is whether the assistant volunteers your name when nobody asked for it by name. Fourth, teams sometimes chase a special AI-only fix — a dedicated llms.txt file, extra schema markup written "for the AI" — expecting it to move the number. Google's own guidance says no special markup or AI-specific file is required for AI Overviews or AI Mode, and an Ahrefs review of roughly 137,000 sites that published an llms.txt file found about 97 percent saw no measurable referral traffic from it. Put that effort into content an assistant can actually cite instead.
Relation to AI recommendations
Mention rate gets you in the room; it is not the sale. Being named and being recommended, or cited as the source behind an answer, are different outcomes, so you read mention rate alongside citation rate and share of voice — your slice of all the providers named for a given topic. If mention rate climbs but leads do not follow, look at how you are named: first choice, or an afterthought in a list of five. It also helps to remember that AI citation is its own selection process, not a mirror of search rankings — an Ahrefs study found only about 6 to 8 percent of URLs cited by ChatGPT also rank in Google's top 10 for the same query, and around 80 percent of ChatGPT's cited URLs do not rank in Google's top 100 at all. In practice you move mention rate the same way you move any Generative Engine Optimization metric: content specific enough to quote, a clear entity the assistant can tie to your brand, and genuine third-party mentions of your name across the web, which the correlation data above suggests matter more than links alone.
Example
A mid-sized bicycle retailer wants to know whether AI assistants know they exist. Their team asks ChatGPT 50 typical customer questions, things like "where can I buy an e-bike with good service in Freiburg" or "which bike shop repairs cargo bikes", running each one three times. The shop's name turns up in 6 of the 50 answers, for a mention rate of 12 percent. After three months of a structured FAQ section, a filled-out Google Business Profile and a repair-advice blog, the team reruns the same 50 questions: now it is 31 percent. Nothing here is a controlled study — it is one shop's own before-and-after tracking — but it is exactly the kind of baseline-then-repeat measurement that makes the metric useful.
Common questions
What is a good mention rate?
There is no universal target, because it depends heavily on category and how crowded it is with competitors. What matters is the trend: a baseline measurement, then regular repeats on the same question set. A rate that is climbing, and that beats your direct competitors on the same questions, matters more than any specific number.
How does the mention rate differ from share of voice?
Mention rate tells you what share of questions name you at all. Share of voice takes it a step further, setting your mentions against every provider named across those same answers, so it shows your relative slice of the conversation rather than just your presence in it. Track mention rate first, then share of voice once you know you are actually showing up.