Interactive guide
AI Visibility Playground
See how ChatGPT, Claude, Gemini, and Perplexity decide which brands to mention, cite, and recommend — and try the same scoring logic your own audit uses.
Most founders assume AI visibility is about ranking higher. It is not. It is about being the clearest, most trusted, easiest-to-recommend answer for a given question. This page is the working reference for how each engine behaves and what to change first.
The four engines
How each one decides
ChatGPT
Indexes via OAI-SearchBot and fetches live pages via ChatGPT-User. Training data is collected separately by GPTBot.
Responds to: Explicit category statements, FAQ content, stated pricing.
Claude
Answers from trained understanding plus material it can retrieve or is given at answer time.
Responds to: Unambiguous positioning and attributable proof — it omits rather than guesses.
Gemini
Leans on Google Search grounding, so ordinary indexation is a prerequisite.
Responds to: Schema.org markup, clean heading hierarchy, strong entity signals.
Perplexity
Retrieval-first — nearly every answer is assembled from pages fetched at query time, with visible citations.
Responds to: Self-contained passages, comparison tables, original numbers.
Try it yourself
Ask the engines about your category
This runs a real category question against every provider you have configured and reports whether your brand comes back — and who gets named instead.
AI answer engine test
Asks each configured engine for the best tools in your category and checks whether they name you.
Score calculator
Estimate your score by hand
Tick what your site already has. The number recalculates live using the same 20/30/30/20 weighting the real scanner applies.
Tick what your site already has
Estimated score
Developing readiness
Same 20/30/30/20 weighting the real scanner uses.
Biggest wins available
- H1 describes what you doStructural Understanding+3.8
- Clear category statement ("X is a Y for Z")Structural Understanding+3.8
- Plain language, low jargonStructural Understanding+3.8
- FAQ section publishedAnswer Selection+3.8
What to fix first
The order that actually works
Access first — if crawlers cannot fetch you, nothing else counts. Then the category statement, because every downstream signal depends on the engine knowing what you are. Then answer-shaped content. Authority work last, since it takes longest and pays least on a site that still cannot be classified.
The full sequenceRelated guides