What Are the Two Doors of AI Visibility?
Short answer
The two doors of AI visibility are the two ways an AI system can know your brand: the training door, what the model learned before its knowledge cutoff, and the retrieval door, what it looks up live when answering. Each door opens through different work on a different timeline.
Two Doors of AI Visibility explained
"Get visible in AI" sounds like one goal, but it covers two separate mechanisms. A brand can be strong through one door and missing from the other, and most strategies only work one of them.
The training door
Models learn about brands from the text they're trained on. If your brand is described consistently across many independent sources, the model learns that association and can name you without looking anything up.
- What opens it: consistent descriptions across the web, third-party coverage, reference listings and reviews (entity footprint)
- Timeline: slow. Changes land only when models retrain, often months later.
- Durability: high. Once learned, an association is hard for a competitor to displace.
The retrieval door
Search-connected AI products fetch live pages to answer questions and cite what they use (RAG).
- What opens it: crawlable pages, direct answers, clear structure and sourced claims (machine-readable content)
- Timeline: fast. A good page can be cited soon after it's indexed.
- Durability: low. You compete for every answer, every time.
AI companies run separate crawlers for each door. OpenAI, for example, uses GPTBot for training data and OAI-SearchBot for ChatGPT search, and site owners can allow or block each one separately.
Reading the symptoms
- Mentioned but rarely cited: the training door is open and the retrieval door is closed. Fix your content and crawl access.
- Cited sometimes but never named unprompted: the retrieval door is open and the training door is closed. Build your footprint.
- Neither: start with retrieval for early wins while the footprint work compounds.
In practice
The framework turns one vague goal into two workstreams with different timelines. Consistency and corroboration feed the training door slowly. Structured, sourced content feeds the retrieval door right away. Most brands I audit haven't invested in either on purpose, and the diagnosis usually points to one door. I named it because I kept drawing the same two-door picture on whiteboards.
Common misconception
AI visibility usually gets treated as one lever, and the lever is usually content. Trained knowledge and live retrieval run on different mechanisms and timelines, so the first step is finding out which door is closed.
Free email course: Smarter SEO for the Age of AI
Seven short lessons over three weeks: why clicks are falling even when rankings hold, and how to get AI systems to recommend your brand.
