Answer engines are a new kind of search

Tools such as ChatGPT, Claude, Gemini and Perplexity, along with AI Overviews inside Google itself, increasingly answer a question directly rather than returning a list of links. A business can lose that moment entirely if its site is unclear, inaccessible to these systems or simply never considered as a source.

This does not replace traditional SEO. It sits alongside it, rewarding many of the same fundamentals: clear structure, genuine expertise and content that actually answers the question asked.

Let AI crawlers in, deliberately

A generic robots.txt wildcard technically allows most AI crawlers through, but it does not say anything about whether you want them there. Naming the crawlers you are comfortable with, such as GPTBot, ClaudeBot, PerplexityBot and Google-Extended, removes ambiguity for both the crawler and anyone reviewing the file later.

An emerging companion file, llms.txt, gives assistants a short, plain-text summary of the business: what it does, who it serves and which pages matter most. It will not force an answer engine to use you, but it removes guesswork that can otherwise lead to a vague or simply wrong description.

Structured data still carries the weight

Schema.org markup that clearly states the business, service, service area, pricing approach and frequently asked questions helps both traditional search and AI systems extract accurate facts rather than infer them from surrounding text.

The same markup that supports a rich result in Google also reduces the chance an AI assistant invents or misattributes a detail about your business, since it has a structured fact to point to instead of a guess.

Write in a way a machine, and a person, can trust

Specific, checkable statements travel further than broad marketing language. Naming the service area, the services actually offered and what is excluded gives an assistant something concrete to quote, rather than something it has to soften or skip.

This also happens to be good writing for people. Visitors trust specifics for the same reason a model prefers them: vague claims are harder to verify and easier to doubt.

Measure what you can, and expect gaps

Most analytics tools do not yet cleanly separate AI-referred visits from direct traffic, so early signals are indirect: branded search volume, direct type-in visits, and enquiries that mention finding you through an AI assistant.

Treat this as an evolving area rather than a solved one. Revisit the basics (crawler access, structured data, clear and specific content) every few months as these platforms and their citation behaviour continue to change.