ChatGPT builds its answers from two main layers: what the model learned during training and, when search is triggered, web pages it reads in real time. Files a user uploads and apps the user connects can add further context.
- Training data: public web text, licensed content and human feedback, frozen at a knowledge cutoff and usually shown without links.
- Web search: current pages found through third-party search providers and OpenAI’s own crawlers, shown with clickable sources.
- User content and connected apps: documents, pasted text and services the user has allowed.
Brands have the most influence over the search layer. OAI-SearchBot crawls pages for ChatGPT search results, ChatGPT-User visits pages on behalf of users and GPTBot collects data for model training; blocking GPTBot is a separate decision from search visibility. Check OpenAI’s current crawler documentation, since names and rules change.
Pages that answer the question in the first paragraph, use clear headings, carry schema.org markup and match what independent sources say about the brand are more likely to be cited. No one can guarantee a position in ChatGPT answers; technical setup typically takes 1–2 weeks, and changes usually surface in AI answers within 4–12 weeks. Racoun’s AI visibility service covers audits, crawler management, structured data and monthly citation tracking.
Talk to Racoun: we are based in Istanbul and work with brands across Türkiye. Message us on WhatsApp or email ekip@racoun.com.
