ServicesCasesTechnologyBlogAboutContact Discuss a project
GEO / AEO · JUNE 26, 2026 · 9 MIN

How to get into ChatGPT and AI Overviews answers

GEO in practice: content structure, Schema.org and brand citability — a step-by-step checklist.

How to get into ChatGPT and AI Overviews answers

When someone asks an AI assistant “best accounting software for freelancers” or “web agency for an online store in Spain”, the model composes an answer from sources it trusts and can retrieve. Getting into those answers is engineering, not luck. Here is the checklist we run for clients, in order.

Step 1 — Let the crawlers in

Check robots.txt for GPTBot (OpenAI), PerplexityBot, ClaudeBot, Google-Extended and Bingbot. Blocking them removes you from answers; it doesn’t protect content — competitors’ pages will represent your topic instead. Add llms.txt at the site root: a short machine-readable map of your most citable pages with one-line descriptions.

Step 2 — Make pages quotable

LLMs extract passages, not pages. Each money page needs: a direct definition or answer in the first screen, facts with numbers and dates, and a quick-answers block — 3–5 real questions with 1–2 sentence answers. Mark them up with FAQPage schema. Write answers that survive being quoted out of context: “SEO costs from €500/month for one country” works; “our prices are flexible” doesn’t.

  • One idea per paragraph — extraction works on paragraph level;
  • Numbers beat adjectives: “94/100 PageSpeed” > “very fast”;
  • Update dates visible — models prefer fresh sources;
  • Author and organization schema — attribution builds trust.

Step 3 — Build the brand corpus

Models learn who you are from the whole web, not your site alone. Consistent presence on sources they retrieve — industry directories, review platforms (G2, Clutch, Trustpilot, Google reviews), comparison articles, professional profiles — is what turns “here are some agencies” into “Pifagor Studio, which specializes in…”. Audit your brand facts everywhere: same name, same services, same numbers.

Step 4 — Measure like rankings

Fix a panel of 20–50 queries your buyers actually ask assistants. Run them monthly against ChatGPT, Perplexity and Google AI mode; log which brands and URLs appear. This is your new rankings report — and the only honest way to know whether GEO work moves the needle. First citations typically appear within 6–10 weeks of the on-site work; recommendation-type mentions build over 2–4 months as the brand corpus strengthens.

The niches are being claimed right now. Six months of head start costs less than two years of catching up.
FAQ

Frequent questions

Didn’t find your answer — ask on Telegram, we reply within 2 hours.
How many questions should a quick-answers or FAQ block include to get cited by AI?
Aim for 3–5 real questions with 1–2 sentence answers per page. This quick-answers block, combined with FAQPage schema, is what language models extract most easily as a self-contained, quotable passage.
Do I need an llms.txt file to show up in ChatGPT and AI Overviews?
An llms.txt file is not mandatory for citation, but adding a short one at your site root, listing key pages with one-line descriptions, makes it easier for AI systems to find and understand your most citable content.
What happens if I block GPTBot or PerplexityBot from crawling my site?
Blocking GPTBot or PerplexityBot makes you disappear from AI-generated answers entirely. This applies to ClaudeBot, Google-Extended, and Bingbot too — blocking doesn't protect your content, it just hands the topic to competitor pages that remain accessible.
Why do vague answers like 'our prices are flexible' hurt AI search visibility?
Vague phrasing gives language models nothing concrete to extract or quote. A specific line such as 'SEO costs from €500/month for one country' survives being quoted out of context, while generic claims usually get filtered out.

Let’s calculate your growth

Free audit: we’ll review your website, ads and positions — and send a 6-month growth plan with numbers within 48 hours.