A business appears in ChatGPT when it (1) is accessible to AI crawlers, (2) offers clearly structured, standalone answer blocks, (3) has correctly implemented schema markup, and (4) is mentioned in trustworthy third-party sources. There is no registration system — AI systems select sources algorithmically.
The good news: not every business needs Google top-10 status to appear in ChatGPT. Content can be cited even from page 2 or 3 — if it's well structured. The bad news: many websites unintentionally block AI crawlers in robots.txt. Then no amount of optimization helps.
These 7 steps cover the most important levers — from technical foundations to third-party source strategy.
7 steps to appear in ChatGPT, Perplexity and Gemini
- Step 1: Allow AI crawlers in robots.txt
GPTBot (OpenAI), PerplexityBot, ClaudeBot (Anthropic), Google-Extended and Bingbot must be explicitly allowed in robots.txt. Blocking even one of these bots means you won't be cited on that platform — regardless of content or page quality. - Step 2: Write standalone answer blocks
AI systems extract passages, not entire pages. Every important piece of content needs to work as a standalone section: ask a question — answer directly — 40 to 60 words — no introduction needed. This paragraph is itself an example of the format. - Step 3: Add an FAQ section with natural questions
Add 4–6 questions that real users ask to every important page. Phrase the answers precisely and completely — not promotionally. AI systems extract FAQ content very often. - Step 4: Implement FAQPage schema markup
Structured data (JSON-LD) helps AI systems classify content. FAQPage schema increases the probability of AI extraction by 30–40%. Also add: Article schema with author, date and publisher. - Step 5: Add statistics and source citations
Concrete numbers with cited sources increase AI visibility by up to +40% (Princeton GEO Study, KDD 2024). Not: "we are the best agency." Instead: "customers reach X on average within Y weeks (internal data, 2025)." - Step 6: Make authorship transparent
Name the author, their qualifications and the company — visible on the page, not just in the legal notice. AI systems prefer content with traceable expertise. E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is a measurable citation factor. - Step 7: Be present in third-party sources
Industry directories (e.g. geoagenturen.de), Google Business Profile, trade magazine mentions — brands are cited 6.5× more often via third-party sources than via their own website. A Wikipedia entry alone accounts for 7.8% of all ChatGPT citations.
How ChatGPT selects sources — the technical background
ChatGPT uses two data sources simultaneously:
| Source | How it works | Optimization approach |
|---|---|---|
| Training data | Information from the dataset up to the training cutoff | Presence in Wikipedia, established trade media, public databases |
| Live search (via Bing) | Real-time web search when browsing is enabled | Bing indexing, structured content, robots.txt for GPTBot |
Perplexity works differently: it always searches the web live and cites its sources with links. That makes Perplexity more accessible than ChatGPT for new or lesser-known websites — because training data plays no role.
Gemini (Google) prefers Google-indexed pages and benefits strongly from classic SEO. If you rank well on Google, you have a head start here.
The most common mistakes in ChatGPT optimization
- AI crawlers blocked: robots.txt blocks GPTBot or PerplexityBot — without anyone realizing it. This often happens from copy-pasting standard robots.txt templates from older SEO guides.
- No schema markup: without FAQPage and Article schema, the page has no structured metadata — AI systems can classify the content less well.
- Promotional language without substance: "We are the leading agency in Düsseldorf" won't get cited. "According to customer project X, we reduced inquiry costs by 34% in 6 weeks" will.
- No author attribution: anonymous content has lower E-E-A-T scores — a measurable disadvantage in AI source selection.
- Answer blocks that are too long: paragraphs over 200 words are less often extracted in full. Better: many short, standalone blocks.
Practical example: what optimized content looks like
Not optimized (typical for many agencies):
"As the leading AI marketing agency in Düsseldorf, we offer you comprehensive solutions for your digital success. Our experienced experts are here for you with cutting-edge technologies." → Don't expect ChatGPT to cite this.
GEO-optimized:
ClientFlow24 – Deutsche Digitale Marketingagentur, Düsseldorf implements GEO/AEO optimization for businesses in the DACH region. Customers achieve measurable AI visibility within 4–8 weeks — with monthly reporting on which queries improved. Contact: +49 160 5900929, Mühligweg 56, 40468 Düsseldorf.
The difference: a concrete name, location, promise with a timeframe, and contact details. Understandable on its own, without context. That is exactly what ChatGPT cites.
Monitoring: how do you know if it's working?
Check these queries manually every month in ChatGPT and Perplexity:
- "Best [your industry] agency in [your city]"
- "What is [your core service]?"
- "[Your service] cost"
- "[Your company name]" directly
Note: Is your company mentioned? Is the right website cited? What phrasing does the AI use? Keep this data in a simple spreadsheet and compare monthly.
For systematic monitoring, we recommend: Otterly AI (Share of AI Voice), Peec AI (multi-platform), or ZipTie (Google AI Overviews + ChatGPT + Perplexity).
Want to make this happen together?
In the free First Look, we check whether your website can be found by ChatGPT and Perplexity today — and show you the fastest path to AI visibility.
Request AI visibility check →