An AI visibility test is a manual check on whether and where your brand shows up in the answers from tools like ChatGPT, Perplexity, and Google AI Overviews. You build 15 to 20 queries that real customers ask, run them across the major AI platforms, and record for each one whether you show up, where, and who is there instead of you. The basic version takes about an hour and no paid tool. This guide walks through the specific process and the common mistakes to avoid.
Treat the result as a directional snapshot for one language, one market, and one moment — not as a definitive measurement of every AI answer.
Why test your AI visibility
AI visibility is a new discipline, and the first step is finding out where you stand — otherwise you’re guessing. Some industry analyses in B2B suggest that a large share of brands are functionally invisible to AI answers, but the results vary widely by market and methodology, so treat them as directional.
The good news: a manual test is cheap and fast. You don’t need a tool, just an hour and a structure. Paid tools only start to make sense for ongoing monitoring, not for a first read.
Step 1: Build the right queries
The foundation of the test is 15 to 20 queries that mirror how your customers actually ask. Split them into four types:
Four query types for an AI visibility test
- Branded queries “What is [brand],” “[brand] reviews,” “experience with [brand].” These check whether AI has the right information about you.
- Category queries “Best [product] in [your city],” “[industry] companies near me.” The most important type — this is where people look for you without knowing your brand.
- Problem-solution queries “How do I fix [problem],” “what can replace [X].” These show whether AI suggests you as a solution.
- Comparison queries “[Brand] versus [competitor],” “alternatives to [X].” These show where you stand against the competition.
Step 2: Test across platforms
Every AI platform cites differently, so testing on one isn’t enough. Three is the minimum.
| Aspekt | Platform | How it cites and what to watch for |
|---|---|---|
| Perplexity | Usually lists sources with links | Very easy to read in a manual test — you can see straight away where the AI is drawing from |
| Google AI Overviews | Shows links inside regular search | Often built from web content and other findable sources; watch which sites appear in the overview |
| ChatGPT | Mentions brands, often without a link | The most widely used; watch whether it describes you correctly |
| Gemini / Claude | Conversational, varying degrees of linking | Supplementary platforms; with Claude, watch whether the answer draws on web sources at all |
Important: test logged out or in a private window. A signed-in account can shift the results through history, memory, or personalization signals. A private window is closer to what an unfamiliar customer sees. Test in your target language and market, and with Google AI Overviews, watch whether an overview appears for that query type at all.
Step 3: Repeat and record
AI answers aren’t the same every time — the same query can return a slightly different answer and different cited sources (technically, they’re non-deterministic). So:
-
Run every query three to five times
A one-shot test skews the picture. Repeating shows whether you appear consistently or only occasionally. For your key queries, more repetitions are worth the time.
-
Record whether AI mentioned you at all
Yes/no for every query and every platform. This is the most basic metric — what percentage of queries mentioned you at all.
-
Where in the answer you show up
Record the first occurrence consistently: 1st / 2nd–3rd / 4th or later / not mentioned. A first mention carries more weight than a mention buried among ten others.
-
Who shows up instead of you
List the brands the AI mentions. If the same competitors keep coming up, you have a clear list of who you're competing with for AI visibility.
-
Which sources the AI cites
Where the AI is drawing from — your site, review platforms, Reddit, Wikipedia? That's the key to your strategy: if AI is describing you from third-party sources, you know where to focus your work on mentions and links from other sites.
Keep your record in a simple spreadsheet: columns for query, platform, appeared (yes/no), position, competitors, sources. After an hour you’ll have an overview that tells you more than any gut feel.
Step 4: Interpret the results
| Aspekt | Test result | What to do about it |
|---|---|---|
| You show up with the right information | The best-case scenario | Keep the content current; watch whether your position slips over time |
| You show up, but the information is wrong | Common with lesser-known brands | Update your own content and structured data; add the facts to findable places |
| A competitor shows up instead of you | A gap against the competition | Analyze what makes the competitor stronger — content, reviews, credibility, and mentions or links from other sites |
| You don't show up at all | Zero AI visibility | Check your robots.txt, your content on the topic, and your mentions off-site |
If you don’t show up at all, work through the causes from the most technical down:
- Are you restricting search crawlers? Check your robots.txt and any site protection (the Block AI Bots toggle in Cloudflare, for example). Bear in mind, though, that blocking alone isn’t the only reason you might not appear in AI answers.
- Do you even have content on the question? AI can’t cite what doesn’t exist.
- Is the brand well known enough? Few mentions off your own site means a weak signal.
- Who is there instead of you, and why? Analyze the competitors showing up.
Automation: when a tool is worth it
A manual test is great for a first audit. For ongoing monitoring, though, a tool pays off — running 20 queries by hand every week isn’t sustainable.
| Aspekt | Manual test (free) | Automated tool (freemium / paid) |
|---|---|---|
| Cost | Free, just your time | Free checks with no signup, plus paid plans for monitoring |
| Best for | A first audit, a small business, a read on where you stand | Ongoing monitoring, multiple brands, reporting |
| Frequency | One-off or occasional | Automatic and regular |
| Scale | Dozens of queries by hand | Hundreds to thousands of queries across platforms |
| Examples | ChatGPT, Perplexity, Google AI Overviews by hand | Free checks and paid monitoring plans (see the list below) |
Specific tools as of June 2026 (features and pricing change over time, so check the current state with the provider):
AI visibility tracking tools (as of June 2026)
- Free, no signup Ahrefs AI Visibility Checker, Semrush AI Search Visibility Checker, Search-visibility.ai — a fast one-off check across platforms.
- Freemium and trials Hall Lite (free plan with limits on projects and questions), Otterly.ai (trial period).
- Paid, for ongoing monitoring Ahrefs Brand Radar, Semrush AI Toolkit, Otterly.ai, Passionfruit Labs — automated tracking, multiple brands, competitive benchmarking.
A practical recommendation: start with the manual test. If you’re working on AI visibility consistently, add a free tool for quick cross-platform checks, and a paid one only once you need regular monitoring and reporting.
The most common testing mistakes
One query, one time
AI answers aren’t the same every time. A single query run once tells you nothing — you need more queries, and each of them repeated. Without repetition you’re testing randomness.
Branded queries only
You already know your own brand. The point of the test is the category and problem queries, where people look for you without knowing your name. Branded queries alone give you a false positive.
ChatGPT only
Every platform cites differently. Testing on ChatGPT alone will miss the fact that Perplexity or Google AI Overviews might not mention you at all (or the other way around). Test at least three.
A signed-in account with history
A signed-in account can shift the results through history, memory, or other personalization signals. A signed-in test shows you a picture shaped by your own past activity. Test logged out or in a private window.
Key takeaways
- A manual AI visibility test takes about an hour in its basic form, and it’s free.
- 15 to 20 queries across four types (branded, category, problem-solution, comparison); the category ones matter most.
- Three platforms at a minimum — ChatGPT, Perplexity, Google AI Overviews.
- Repeat three to five times, because AI answers aren’t the same every time, and record appearance, position, competitors, and sources.
- Test logged out for an objective picture.
- An automated tool makes sense for ongoing monitoring, not for a first audit.
If the test shows that AI isn’t citing you, the next steps — content, structured data, mentions off your own site — are exactly what an AI SEO audit works through.
Would you rather not run the test by hand? This site is run by Sniper Design — an AI SEO audit walks your visibility across ChatGPT, Perplexity, and Google AI Overviews, compares you against your competitors, and lays out a specific plan of priorities.
Tool names and pricing in this article come from publicly available information as of June 2026; specific features and plans change over time, so check the current state with the provider. Sniper Design runs AI audits with real cross-platform testing, not just a single tool.
