Why this matters now

Buyers now form first impressions inside a chat window. 6sense found 94% of B2B buyers used large language models during their buying process in 2025. On the consumer side, 68.5% of AI shoppers in Exploding Topics' April 2026 survey used AI mainly for product research.

The industry calls this work answer engine optimisation or generative engine optimisation, AEO or GEO. It overlaps with SEO and rewards different writing. Search ranks pages. Assistants quote passages.

What AI assistants quote

  • A clean definition in one sentence. Subject, verb, meaning, placed early.
  • Answers directly under their questions. The heading asks, the first sentence answers in full, detail follows.
  • Attributed numbers. A figure with a named source and a year survives being quoted. A bare percentage gets dropped.
  • Self-contained paragraphs. Each one stays true when lifted out on its own.
  • Everyday nouns. Use the words your buyer types. Invented category names break the match between question and answer.

What to publish

Definitions of the terms your category uses

Definition questions are among the most common, and most categories have thin answers. The clearest definition available is often the highest-return page you can write.

Fair comparisons

Buyers ask assistants to compare options all the time. A comparison that admits where an alternative wins reads as credible and gets cited. One where your product wins every row reads as an ad.

Your own numbers

Assistants cite sources for facts. Publish data you measured yourself and you become the primary source everyone else repeats.

Clear answers to awkward questions

Price, limits, who the product suits less well. Buyers ask these in every evaluation. When your site stays silent, a third party answers for you.

The technical checklist

  • Allow the crawlers that let assistants answer from your pages: OAI-SearchBot for ChatGPT search, PerplexityBot for Perplexity, and Claude-SearchBot with Claude-User for Claude. GPTBot and ClaudeBot are training crawlers, which is a separate decision.
  • Google's AI Overviews and AI Mode follow Googlebot and the usual snippet controls. Google-Extended covers Gemini Apps and Vertex AI grounding, and Google states it leaves inclusion in Search unchanged.
  • Put your content in the page's initial HTML. Crawlers that skip JavaScript otherwise see an empty page.
  • Add FAQ, Article and Organization schema, so each question maps clearly to its answer.
  • Keep one clear topic per page. A page about six subjects matches all six weakly.

How to check whether it works

  1. Ask the assistants yourself. Put the 20 questions a buyer would ask to ChatGPT, Gemini and Perplexity every month, and record which companies get named. It is slow, and it is the most reliable signal available.
  2. Watch referral traffic. Visits from chat interfaces appear in analytics as their own referrers.
  3. Watch branded search. A recommendation in a chat window often leads to a search for the name a few days later.

What this looks like in practice

For one enterprise software company, the work followed a simple sequence. Find the terms buyers type. Audit what the existing answers leave out. Write one page that answers the core question in its first sentence, then supporting pages that each answer one nearby question completely. Mark it all up. Then wait while indexing catches up.

In about two months, ChatGPT and Gemini were citing the company for its category term. Positions in AI answers move, so they need regular re-checking.

Good writing and AI visibility are the same job

The passages that get quoted are clear, specific, sourced and willing to concede a point. That describes good writing, and always did.