AI Visibility Is a Measurement Problem Before It Is an Optimization Problem
There is no universal number one position in ChatGPT.
Different AI systems use different search, retrieval, citation and recommendation processes.
That means useful AI search and GEO work begins by establishing how the brand appears today.
What I test
- Priority customer questions.
- Brand and competitor inclusion.
- Cited sources.
- Uncited mentions.
- Recommendation context.
- Incorrect facts.
- Prompt language.
- Country and location differences.
- Repeatability across sessions.
- Whether live search or retrieval appears to be involved.
Platforms
Depending on the client and audience, testing can include ChatGPT, Google AI, Gemini, Perplexity, Grok, Claude and Microsoft Copilot.
Not every platform deserves equal attention for every business.
What can be improved
- Crawler and retrieval access to the website.
- Service and product information.
- Entity consistency.
- Source quality.
- Content gaps.
- Third-party corroboration.
- International coverage.
- Structured data where it serves an established purpose.
- Product feeds for applicable e-commerce systems.
- Reputation sources.
What cannot be guaranteed
- A fixed AI ranking.
- A recommendation.
- A particular sentence.
- A citation.
- A fixed appearance date.
- Identical answers for every user.
How this differs from SEO
Much of the foundation is standard SEO.
The additional work involves prompt-based testing, citation mapping, cross-platform comparison, answer monitoring and analysis of the external sources shaping generated responses.