Is your company ready to appear in AI search? A 20-point self-check

The checklist has five areas and 20 items in total. The line after each item is why we look at it.

1. Discoverability

• Are the main pages crawlable? — if robots.txt blocks the search-side AI crawlers (OAI-SearchBot, PerplexityBot and others), every item after this is meaningless. Bot list and configuration
• Is the core content indexed? — check the Bing index as well as Google. Bing is part of ChatGPT’s retrieval path. The Bing index we missed ourselves
• Are the CDN or firewall blocking bots? — robots.txt open while a security appliance blocks does genuinely happen
• Does the body text exist without JavaScript? — many AI crawlers do not execute scripts. Open the page with scripts disabled and check whether the body is visible

2. Entity

• Is the company name consistent across every channel? — a scattered legal name, brand name and English name can be recognised as separate entities
• Is “what kind of company this is” defined in one sentence? — that sentence has to be identical on the home page, the about page and in the structured data
• Is entity information such as representative, address and contact published? — a brand whose existence cannot be confirmed slips down the recommendation candidates
• Is the same information confirmable on the external web (company databases, press)? — divergence between your site and outside information costs trust

3. Content

• Is there a document that answers what customers actually ask? — a site with only a company introduction is not the answer to any question
• Does the answer complete near the top of the document? — AI cites paragraphs. A conclusion at the end cannot be cut out. The structure of an article that gets cited
• Is there a definition, data or experience that exists nowhere else? — recycled content is a type Google’s own guidance explicitly excludes. Our own field data
• Do the claims carry evidence and sources? — a figure with no source gives neither AI nor a person a reason to cite it

4. External trust

• Is the company mentioned outside its own site? — mentions from independent sources such as press, industry material and partner pages
• Is the area of expertise confirmable externally in the same wording? — the entity becomes distinct when “what kind of company” agrees with outside material
• Is the author stated on the content? — a document where the writer is unknown is hard to evaluate for trust
• Are verifiable cases and records published? — specific records with conditions attached, not “many success stories”

5. Measurement

• Is a core question set defined? — with no list of questions to measure against, there is no way to judge improvement
• Are brand mentions and source citations tracked separately? — the two metrics move independently
• Do you look at share against competitors? — your own figures can hold steady while your relative position falls as competitors rise
• Do you re-measure the same questions regularly? — answer engine results keep changing. A single check is one sample

After the check — a prescription per area

The prescription depends on which area is empty. If discoverability is empty, technical work comes first (the order of diagnosis); if content is empty, documents in question-and-direct-answer form have to accumulate; and if measurement is empty, the baseline has to be recorded before anything else — start without one and there is no way to judge improvement later.

If the technical items are hard to confirm yourself, or you want all 20 checked against actual data, a free AEO audit covers them alongside your current AI visibility for the main questions.