GEO AI visibility · Generative Engine Optimization
Measuring AI visibility: does ChatGPT mention your brand when customers ask for recommendations?
Every week Rivly puts your customers' buying questions to ChatGPT, Perplexity, Claude and Gemini and shows who is mentioned: you or your competitors. With position, cited sources and blind spots.
In short
Measuring AI visibility means putting your customers’ buying questions to ChatGPT, Perplexity, Claude and Gemini and counting how often and at which position your brand is named, compared with competitors. Because answers fluctuate, only a repeated measurement counts. By hand a first test takes about 30 minutes; with Rivly it runs every week.
What you see · demo example alpenkiste.de, fictional brands
Four numbers that are created anew with every measurement. Each with context and comparison.
Mention rate
31 %
of the measured AI answers mention you
First choice
4×
Answers in which you are mentioned first
Position when mentioned
Ø 2.4
Bergwerk Outdoor: Ø 1.6
Blind spots
3
Buying questions in which only competitors appear
| Buying question | ChatGPT | Perplexity | Claude | Gemini |
|---|---|---|---|---|
| Online shops for hiking gear | Position 3 | Position 1 | – | Position 2 |
| Down jacket under 300 euros | – | – | Position 2 | – |
| Trekking rucksack with advice | Position 1 | Position 1 | Position 2 | Position 1 |
| Alpenkiste or Bergwerk Outdoor | Position 2 | Position 1 | Position 2 | Position 2 |
| Breathable rain jacket review | – | Position 4 | – | not read |
How Rivly works in this channel
Buying questions instead of keywords
12 to 20 questions the way your customers ask them: “Which shops for hiking gear can you recommend?” Rivly suggests them from your category, you change whatever does not fit.
Four systems, one table
Every question goes to ChatGPT, Perplexity, Claude and Gemini. Each cell shows: mentioned, position, not mentioned. Plus the complete answer text.
The sources the AI systems cite
Which domains the systems name as evidence. Whoever is missing there is usually missing from the answer as well. Your recommendations grow out of this.
Method
How AI visibility is measured reliably
A single question to ChatGPT says little about whether customers find your brand. Five building blocks make the measurement reliable. If one is missing, the number is a chance find.
1. Buying questions, not keywords
You measure with full questions as a prospect asks them who does not know your brand yet: “Which agency for B2B marketing in Munich can you recommend?” Your own brand name stays out of the question. For most brands 15 to 20 questions on finding, comparing and deciding are enough.
2. Four AI systems
ChatGPT, Perplexity, Claude and Gemini answer differently and use different sources. Only all four together show whether your brand appears broadly or in one system only.
3. Repeated, not once
The same question leads to different answers on the same day. In Rivly’s test runs, for about one question in five the mention of the brand flipped (19 % of 64 questions). A number is a measurement only when it comes from the same questions week after week.
4. Against competitors
A 30 % mention rate is strong if the strongest competitor sits at 15 % and weak if it sits at 60 %. So the same questions are evaluated for your brand and for 3 to 5 competitors.
5. With the cited sources
Perplexity and ChatGPT with web search name the pages they rely on for many answers. These sources show where your competitors appear and you do not. That is where the work starts.
Key figures
The four numbers a buyer reads
A few key figures come out of the answers. Each answers a question you can explain to a managing director in one sentence.
| Key figure | What it shows | How to read it |
|---|---|---|
| Mention rate | Share of measured AI answers that name your brand. | The most important single value. Compare it with the strongest competitor and over several weeks, not with one answer. |
| First mention | Answers in which your brand stands ahead of all others. | The first position weighs more than a later one. A high mention rate with few first mentions means: you are in, but not recommended. |
| Share of voice in AI answers | Your share of all mentions in the competitive field. | Shows whether you gain or lose room in the field, even when your own rate stays the same. |
| Cited sources | Domains the systems name as evidence. | Whoever is missing there is usually missing from the answer. The gaps become concrete tasks. |
| Blind spot | A buying question where only competitors appear. | The questions with the largest expected effect. Start there. |
Self-test
Check it yourself in 30 minutes
Without a tool, a first impression takes about 30 minutes. You need a spreadsheet and four browser windows.
- Write down 5 buying questions without your brand name. Example: “Which consultancy for supply chain audits in mid-sized companies can you recommend?”
- Open ChatGPT, Perplexity, Claude and Gemini in a private window without signing in, so earlier chats do not colour the answer.
- Ask every question in every system. That is 5 questions times 4 systems, 20 answers, at roughly a minute and a half each.
- Note for each answer: is your brand named, at which position, and which competitors stand ahead.
- Divide the number of answers naming your brand by 20. That is your mention rate for this week.
- Repeat the test a week later with the same questions. Only the comparison shows a direction.
The self-test gives a picture of 20 answers in one week. For a trend, competitors and sources you would have to repeat it every week with all questions. The check does that: enter your domain, create a trial account, and the first measurement is ready after about 30 minutes. You get 14 days of free trial access, no credit card needed. A result without any account does not exist, because Rivly first has to recognise and query your brand and competitors.
Own measurement
What the Rivly study shows about AI visibility
For the AI study 2026 Rivly put 80 buying questions to four AI systems, measured on 27 September and on 5 and 18 October 2026. Two results matter for measuring.
- Google rank helps little: only 42 of 872 brand mentions (4.8 %) belonged to brands that stood on page 1 of Google with their own domain for the same question. Measure AI visibility separately from Google.
- The top moves slowly: from 5 to 18 October the systems named the same brand first in 201 of 303 pairs of question and AI system (66 %). A weekly measurement shows change; a daily one would be noise.
Tool or by hand
What a tool costs and when it pays off
Measuring by hand costs only time: around 30 minutes per test, and without competitor comparison and trend the result stays thin. A tool pays off as soon as you want to know regularly how your mentions develop against competitors.
After the 14-day trial Rivly costs 69 € per month in the One plan (1 project, 3 competitors, 15 buying questions) and 179 € per month in the Growth plan (3 projects, 5 competitors, 20 buying questions). Agencies with up to 10 client projects choose Agency for 449 € per month. All plans can be cancelled monthly.
Definition
In short: what is AI visibility?
AI visibility describes whether and in which position a brand appears in the answers of AI systems such as ChatGPT, Perplexity, Claude and Gemini when people ask for products, suppliers or recommendations. The work on it is called Generative Engine Optimization (GEO). Rivly measures the result: every week, with fixed buying questions, in comparison with your competitors.
Other names for the same thing
Every industry and every tool calls it something else. The meaning is the same, and Rivly measures it.
- Buying-question tracking
- Share of voice in AI answers
- Check ChatGPT visibility
- AI answer monitoring
- AEO, answer engine optimisation
- AI visibility check
Frequently asked questions
About AI visibility
How do you measure AI visibility?
With fixed buying questions you put to ChatGPT, Perplexity, Claude and Gemini every week. You count in how many answers your brand appears (mention rate), at which position, and which competitors and sources are named. A single answer is not a measurement, because the same question leads to different answers.
Can I check AI visibility for free?
Yes, in two ways. By hand it only costs time: 5 buying questions in four AI systems, about 30 minutes. With Rivly you get 14 days of free trial access without a credit card and the first measurement after about 30 minutes. Rivly offers no result without an account.
What does an AI visibility tool cost?
After the 14-day trial Rivly costs 69 € per month (One, 1 project, 15 buying questions), 179 € per month (Growth, 3 projects, 20 buying questions) and 449 € per month (Agency, 10 projects). All plans can be cancelled monthly.
How often should you measure AI visibility?
Once a week with the same questions. Answers fluctuate from day to day, so only the trend over several weeks shows a direction.
Is a good Google ranking enough for AI visibility?
No. In the AI study 2026 only 42 of 872 brand mentions (4.8 %) belonged to brands standing on page 1 of Google with their own domain. Measure Google position and AI mention separately.
Why four AI systems and not only ChatGPT?
The systems answer differently. Gemini leans heavily on Google data, Perplexity cites current sources, ChatGPT often lists several suppliers. A brand can lead in one system and be absent from another. Only the comparison shows where work is needed.
AI answers fluctuate. How does that become a reliable number?
Rivly measures every week with the same buying questions and shows the trend. A single answer is not a measurement, the series over weeks is. If an answer fails for technical reasons it does not count as a loss, the denominator gets smaller.
What is a blind spot?
A buying question where at least one competitor is mentioned and you appear in none of the systems. Blind spots are the recommendations with the largest expected effect.
How can I check the AI visibility of my brand?
Two ways. Manually: ask ChatGPT and Perplexity the questions your customers would ask, for example “Which suppliers of hiking gear can you recommend?”, and note over several weeks which brands are mentioned. Automatically: Rivly puts your buying questions to ChatGPT, Perplexity, Claude and Gemini every week and shows your competitors next to your brand. First results in about 30 minutes, 14 days free.
Read on in the guide
The other three channels