Our services

If you can think it, we can make it brainsoft.

GEO and AEO: optimising for AI answers, not just blue links

Written By: BrainSoft In SEO

A client asked us why their traffic from search was flat while their competitors kept showing up in ChatGPT and Google's AI overviews. The pages ranked fine. The problem was that nothing on the site was written or marked up in a way an answer engine could lift cleanly, so the models summarised around them.

GEO (generative engine optimisation) and AEO (answer engine optimisation) are the work of making your content easy for AI systems to retrieve, parse and quote. In practice that means clear question-shaped headings, direct answers near the top, structured data, and crawler access for AI user agents. Here is the process we follow.

The steps we actually run

  1. Audit what the models already say about you. Ask the main assistants the questions your customers ask. Save the answers. You want to know which pages get cited, which get paraphrased wrong, and which competitors appear instead of you. That gap list drives everything else.
  2. Rewrite headings as real questions. "Our approach" tells a retrieval system nothing. "How much does a DevOps migration cost?" maps directly onto a query. One question per section, answered in the first two sentences underneath.
  3. Put a short direct answer under each heading. Forty to sixty words, plain declarative sentences, no preamble. Models chunk pages by section, so the answer needs to survive being quoted alone.
  4. Add structured data that matches the visible text. FAQPage, HowTo, Article, Organization. The markup has to describe what is on the page. Mismatched schema is worse than none.
  5. Check crawler access. Look at robots.txt and any WAF or bot rules. Several AI crawlers identify themselves by user agent, and blocking them by accident is common after a security hardening pass.
# robots.txt - allow the answer-engine crawlers you want
User-agent: GPTBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

Decide this deliberately. If you want citations, let the crawlers in. If you have licensing concerns, block them and accept the trade-off. Just do not leave it to whatever a plugin shipped by default.

  1. Keep facts in HTML, not only in JavaScript. Answer engines mostly read the rendered or raw HTML. Prices, dates, addresses and specs buried behind client-side rendering often get missed. Server-render the parts you want quoted.
  2. Add a machine-readable summary. A concise /llms.txt at the root listing your key pages with one-line descriptions helps agents find the right document faster. It is not a standard yet, but it costs ten minutes.
  3. Measure citations, not just rankings. Track which URLs get referenced in AI answers for your target questions, and re-run the audit monthly. Rankings in classic search tell you almost nothing about this.

Most of this overlaps with good technical SEO, which is convenient. If you need help with the crawler, rendering or schema side, our services cover it.

Why plain answers beat polished copy

Marketing prose is written to be read in sequence. Answer engines pull fragments. A paragraph that starts with three sentences of context before the point gets truncated or skipped. We have had better results cutting the warm-up and leading with the fact, then explaining. It reads a bit blunt to humans at first, and then they stop complaining because they find what they need faster.

What does not work

  • Keyword stuffing aimed at prompts. Retrieval is semantic; padding hurts.
  • Schema markup that claims FAQs the page does not show.
  • Hiding text for bots. It gets ignored or penalised.
  • Assuming one assistant's behaviour generalises. Test each one you care about.

Frequently asked questions

Is GEO different from SEO?

It overlaps heavily. Both need crawlable HTML, clear structure and fast pages. GEO adds emphasis on self-contained answer blocks, question-shaped headings, and tracking citations inside AI answers rather than positions on a results page.

Do I need to block or allow AI crawlers?

That is a business decision. Allowing them gets you cited in answers. Blocking them protects content from training use but removes you from those answers. Check your robots.txt and firewall rules, because security tooling often blocks them silently.

How do I know if it is working?

Run the same set of customer questions through the assistants you care about every month and record which URLs get cited and whether the facts are right. That log is more useful than ranking reports for this kind of work.


#SEO