AI response accuracy
Is what AIs say about your brand factually true?
Open the page in the appThe principle
When a prospect asks ChatGPT “what is [your brand]?”, the answer can contain errors: wrong industry, invented services, wrong location, outdated information. The AI Accuracy tab of the GEO Lab queries the 4 engines (ChatGPT, Perplexity, Gemini, Claude) with a neutral question about your brand, then fact-checks each answer against your official website (the reference). It is the factual complement to the Perception page: Perception analyzes the tone AIs use about you, Accuracy verifies the facts.
How to use it
- 1. Open the AI Accuracy tab of the GEO Lab (AI response accuracy page).
- 2. The analysis starts automatically: the 4 engines are queried, then fact-checked (allow 30 to 60 seconds).
- 3. Start with the average score /100 and the global summary — the exact question asked to the engines is displayed just below.
- 4. Go through each engine's card: score /100, summary, list of correct facts (✓) and inaccuracies (✗) — each inaccuracy quotes the engine's wrong claim and its correction (“Reality: …”), color-coded by severity.
- 5. Click “See the engine's full response” to read each AI's raw text.
- 6. After fixing your online presence, click Re-check to run the analysis again and measure progress.
Priority fixes
Under the global score, the Priority fixes box lists the concrete actions to take first to correct the most widespread inaccuracies (e.g. clarify your offer on the homepage, update a piece of information that several engines got wrong).
When an engine doesn't know the brand
If an engine barely knows your brand, its card says so explicitly: the priority is then not to correct facts, but to build presence — mentions, entities (Wikidata, Organization schema), citable content. The Entity & brand and GEO Audit pages are your next steps.
History
Every check is recorded: from 2 readings onward, an Accuracy over time chart plots the average score — hard proof that your fixes are paying off.