AI Search Optimization (GEO)
AI search optimisation, or GEO, makes a site usable as a source by AI answer engines like Google AI Overviews, ChatGPT, Perplexity and Copilot. It requires crawler access without JavaScript, self-contained passages that fully answer a question, and unambiguous entity information the model can attribute to you.
Prices shown are starting points. The final cost depends on the size and scope of your project, and we confirm it in writing before any work begins.
AI engines answer without sending a click. If your content is the source, you get the citation and the credibility. If not, your competitor does, and you never see the query in any analytics report because there was no visit.
Two things decide whether you are usable. First, access: most AI crawlers do not execute JavaScript, so a client-rendered site simply is not readable to them regardless of content quality. Second, structure: models lift passages that answer a question completely on their own, so content that only makes sense after three preceding paragraphs rarely gets cited.
What's included
- AI crawler access audit: GPTBot, PerplexityBot, ClaudeBot, OAI-SearchBot
- Server-side rendering or prerendering verification
- robots.txt review for AI user agents
- Direct-answer passage structure across key pages
- FAQ and question-based content targeting real queries
- Entity clarity: consistent naming, schema, about and author signals
- Citation tracking across AI platforms
- Comparison and definition content that models reach for
Questions people actually ask
Is AI search optimisation different from normal SEO?
The foundations overlap: crawlability, structure, authority. The differences are real though: AI engines favour self-contained passages, explicit definitions and clear comparisons, and they weight consistency of entity information more heavily than traditional rankings do.
Should I block AI crawlers?
It depends on your goal. Blocking protects content from training use but removes you from AI answers entirely. For a services business trying to be found, being citable is usually worth far more than the training-data concern. For a publisher with licensable archives, the calculation differs.
Do I need an llms.txt file?
It costs almost nothing to add, but be realistic: it is a proposed convention with limited confirmed adoption, and Google has said it does not use it for Search. Server-rendered HTML and clear structure do far more work. Anyone selling llms.txt as the main deliverable is overselling.
How do I know if AI engines are citing me?
You have to check directly: query the platforms for your target questions and record whether you appear. Referral traffic from ChatGPT and Perplexity shows in analytics, but AI Overviews citations often produce no click at all, so analytics alone will understate it.
Does my React or Vue site need rebuilding for this?
Not rebuilding: prerendering. A client-rendered site can be statically generated or server-rendered so crawlers get real HTML while users keep the interactive experience. That is exactly what we did to this site.