AiVisible
Practical guide

How to measure AI Visibility: tools, methods and metrics for B2B

You cannot improve what you do not measure. Which indicators to track, which tools to use and how to build a simple system for monitoring visibility in AI.

AI Visibility

article topic

7

sections to work through

25 March 2026

last updated

AI platform features change quickly. This article describes what was verified on the date of the last update. Check the provider's current documentation before making an implementation decision.

Measuring AI Visibility is one of the hardest aspects of this strategy - and one of the most important. Without measurement you do not know whether the work is producing effects or wasting time. The difficulty is that AI has no transparent visibility API: every query can produce a different answer, and measurement tooling is only now maturing.

A useful measurement system can be built without expensive tooling. It requires consistency, complete records and stable rules. Below is a framework aligned with the AiVisible methodology.

The core indicator: presence rate

The primary AI Visibility KPI is the presence rate - the percentage of test queries in which your company appears in the answer. To calculate it: choose a set of 20 to 30 queries a buyer might ask. Test each three times in a given AI system. Count how many times your company appeared, divided by all possible appearances.

An example: 25 queries x 3 repetitions = 75 opportunities in ChatGPT. The company appeared in 18 cases, a presence rate of 24%. The same measurement a month later: 31 cases, or 41%. A rise of 17 percentage points - concrete, measurable progress.

Four dimensions of AI Visibility measurement

DimensionWhat it measuresHow to measure itFrequency
PresenceWhether the company appears in answersManual query testing x 3Monthly
PositionWhere in the answer the company appearsNote the position in the answer (1-5)Monthly
ContextHow AI describes the companyQualitative analysis of the answersQuarterly
Model coverageHow many models it is visible inTest in ChatGPT, Perplexity, Gemini, Google AIMonthly

How to build a test question set

The question set is the heart of the measurement system. It has to be stable (the same questions each month so results are comparable), representative (reflecting real buyer questions) and varied (different intents and levels of detail).

  • General recommendation questions: "Which [your category] companies do you recommend?"
  • Niche recommendation questions: "Which [your category] company specialises in [your niche]?"
  • Location questions: "A good [your category] company in [your city or region]"
  • Problem questions: "How do I solve [the problem you solve]?"
  • Comparison questions: "How do I choose a [your category] supplier - what should I look at?"

Tools for measuring AI Visibility

The market for AI Visibility monitoring tools is young and changing fast. A summary by category:

ToolTypeWhat it doesCost
Manual testing (ChatGPT, Perplexity)ManualBasic presence measurementFree
Brandwatch / MentionMention monitoringTracks brand mentions across the webFrom around USD 50/month
Ahrefs Brand RadarMention monitoringBrand mentions in AI answers and LLMsFree plan available
Semrush AI ToolkitSEO plus AITracks AI Overviews for keywordsFrom around USD 99/month
Peec.AI / Otterly.AIAI monitoringAIO triggers and LLM trackingFrom around USD 29/month
An analytical serviceSpecialistA Snapshot or audit with interpretationDepends on scope

A monthly measurement protocol, step by step

  1. 1Day 1: open a new ChatGPT session with no history. Test all 25 questions, each three times. Record the results in a sheet: query, model, repetition, appearance (yes or no), position (1-5) and description (how AI characterised the company).
  2. 2Day 2: the same in Perplexity. Additionally, record which pages Perplexity cites as sources for your queries.
  3. 3Day 3: Google AI Overviews. Enter the queries in Google and check whether an AI Overview appears and whether your page is cited.
  4. 4Days 4-5: calculate the indicators. Presence rate per model, average position and comparison with the previous month.
  5. 5Day 5: check external mentions. How many new mentions of the company appeared in the past month?

How to interpret the results - a benchmark

Presence rateAssessmentWhat to do
0-10%Critically lowA full AI Visibility strategy from the ground up
10-25%LowContent optimisation plus mention-building
25-50%AverageTargeted work on the weaker areas
50-70%GoodMonitoring, maintenance and expansion into new niches
Over 70%Very goodDefend the position and analyse the quality of the descriptions

The benchmark depends on the category. In very narrow niches, 30% can mean market dominance. In popular categories, even 60% is only a good position. Always measure against direct competitors, not against an absolute value.

Measurement traps to avoid

  • Testing in the same session - AI retains context. Always start a new session so earlier answers do not influence the results.
  • Too small a question set - five queries is not enough for a meaningful result. The minimum is 20, ideally 30.
  • No repetitions - asking a query once is not a measurement. Ask each at least three times to average out the model's randomness.
  • Comparing results across different model versions - ChatGPT after an update can give different results not because your company is more visible, but because the model changed.
  • Measuring only your own company - a number without context says little. Always measure two or three competitors too.
Next step

Assess whether your category is suitable for AI Visibility measurement.

We assess category fit free of charge. Full visibility and competitor measurement is delivered within a Snapshot or a strategic audit.

0 PLN · Fit assessment, without a promise of a full report

Review engagement stages

Frequently asked questions

What is the most important AI Visibility indicator?

Presence rate - the percentage of answers to key category queries in which your brand appears.

How often should AI visibility be measured?

Given how quickly models change, we recommend a full measurement monthly, to track trends and the effects of changes.