AI visibility benchmark by industry: what is normal?

An AI visibility score is only usable if you look at it industry by industry. A mention frequency that is excellent in one sector can be below par in another, because the number of available sources, the kind of questions buyers ask and the competitive pressure differ enormously from one industry to the next. This article explains why “what is normal?” has no universal answer, which factors determine your score per sector, and how to build an honest benchmark for your own industry instead of staring at a bare number.

What is an AI visibility benchmark?

An AI visibility benchmark is a reference point that shows how often and how accurately AI systems such as ChatGPT, Perplexity, Google AI Overviews or Gemini name your brand, measured against what is realistic in your context. The core of it: “often enough” is not an absolute number. It is the extent to which you appear compared with your direct competitors on the same purchase-driven questions.

Most tools give you a percentage: in so many percent of the prompts tested, your brand gets mentioned. That percentage feels objective, but it is meaningless without context. The same score can mean you are the market leader in a niche sector or a laggard in a crowded category. That is why every serious measurement starts with the question: compared with what? Which mention exactly counts, and why a mention weighs more than a backlink, you can read in our explainer on brand mentions over backlinks.

Why does AI visibility differ so much by industry?

AI visibility differs by industry because language models draw on the source landscape of that sector, and that landscape has taken shape differently in every market. Three factors explain most of the differences.

The source landscape. AI models build their picture of a market from the sources that exist about that market: trade media, comparison sites, review platforms, forums, reference works and company registers. In a software category there are often dozens of independent comparisons and review sources, which gives models rich material to name and position brands. In a traditional industrial or local services sector that material is far thinner, so models answer more cautiously or more vaguely. How AI decides which sources it trusts, we explain in grounding: how AI determines what is true.

The type of questions buyers ask. AI answers appear far more often on informational and orientation questions than on purely transactional searches. An industry in which buyers mainly ask for explanations, comparisons and “what is the best” questions simply generates more moments where a brand can be named. An industry in which people mainly look up a specific supplier they already know offers fewer of those moments. So your benchmark partly depends on how your audience asks questions in the first place.

The maturity of the competition. In some sectors, brands have been building their presence beyond their own site for years, which sets the bar high. In others, almost nobody has worked on AI visibility structurally, so a modest effort already puts you well ahead. The same score therefore means something entirely different depending on how far your competitors already are.

What is a normal score in my industry?

There is no universal “normal” percentage, and you should distrust any number that suggests there is. The only reliable norm is the score of the brands your buyers compare you with, measured at the same moment with the same questions.

We deliberately do not give sector averages in figures, for two reasons. First, the landscape changes fast: models get updated, AI answers are rolled out more widely and the share of questions with an AI answer keeps shifting. An “industry average” from today is out of date next quarter. Second, the benchmark percentages doing the rounds are almost never based on your exact questions, language and region. A figure from an English-language B2C study says nothing about a Dutch-language B2B niche market. Better an honest “it depends on your context” than a falsely precise number that steers you the wrong way. That emphasis on leads and correct mentions over vanity figures runs as a common thread through our entire approach to Generative Engine Optimization.

How do I build a benchmark for my sector myself?

You build a usable benchmark by mapping your own sector at a single point of measurement and then tracking your progress over time. In practice that comes down to a few steps.

1. Draw up a fixed list of purchase-driven prompts. Start from the questions your buyers really ask in their own words, including category, region and target group (“best [category] for [type of company] in Belgium”). Not abstract keywords, but full questions the way a human types them into AI.

2. Determine your reference set. Pick three to five competitors that buyers realistically compare you with. They form your benchmark, not an anonymous sector average.

3. Test across multiple models. Ask the same questions of the major systems, because they draw on different sources and give different answers. How to approach that per platform, you can read in our guides on getting found in ChatGPT and getting found in Google AI Overviews.

4. Note not just whether, but how. Are you named, in which position in the answer, and is the description of what you do correct? A mention in the wrong category is not a win.

5. Repeat at fixed moments. The most valuable benchmark is your own trend line. Measure every quarter with the same prompts, so you see movement instead of a snapshot.

Which measurement instruments fit here and what they can and cannot do, we discuss in our overview of AI visibility tools. And which units of measurement to track per indicator, you will find in the 5 core indicators of AI visibility.

Which score really counts for B2B?

For B2B, what counts is not your average across all questions, but whether AI names you correctly on the handful of questions where a buyer makes a choice. A high total percentage that mainly comes from irrelevant or informational questions is misleading if you are precisely absent at the moments that lead to a quote request.

B2B markets are moreover often narrow, local and specialised. That means fewer sources, less noise, but also fewer questions where you can appear. In such a context a lower absolute percentage is entirely normal, and the right question is whether you appear on the specific, purchase-ready prompts of your audience. Why B2B demands its own approach, we set out in GEO for B2B. The practical lesson: do not chase a pretty number, but presence at the right moments. That is also how we steer, on leads and correct mentions, not on vanity scores.

The short summary

An AI visibility benchmark only makes sense per industry, because mention frequency is determined by the source landscape, the type of questions and the maturity of your competitors. There is no universal “normal” number. The honest approach is to measure your own sector at a single moment against your real competitors, and above all to track your own progress over time, with a focus on the purchase-ready questions that lead to leads. Be sceptical of sector percentages doing the rounds: they are rarely based on your questions, language and region.

Want to know what is normal in your industry and where your brand stands today in AI answers? Take a look at our AI search optimisation or schedule your free intake. Then we will set up a benchmark together that does fit your market.