AI Visibility
How to get your business cited by ChatGPT and Perplexity: a 12-point checklist
A concrete checklist we run on every AI Visibility audit. Each item is something you can verify on your own site in minutes, ordered by how often it is the thing holding a site back.
This is the checklist we use when a client asks why their competitor shows up in AI answers and they don't. Work through it top to bottom. The early items are cheap and are the most common blockers; the later ones are slower and compound.
Access: can the engines even reach you?
1. AI crawlers are allowed in robots.txt
Open yourdomain.com/robots.txt. You want explicit Allow: / rules for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended, or at minimum no Disallow aimed at them. Also check your CDN or firewall. Cloudflare and similar services can block "AI bots" with a single toggle, and that toggle is often on without anyone remembering why.
2. Your pages render without JavaScript
Most AI crawlers fetch HTML and do not execute scripts. If your service descriptions only appear after a client-side framework renders them, the crawler sees an empty shell. Disable JavaScript in your browser and reload your key pages; whatever is missing is invisible to the engine.
3. A sitemap exists and is listed in robots.txt
Engines discover deep pages through the sitemap. Case-study and FAQ pages, which are the most citable content you have, are exactly the pages that tend to be missing from it.
Identity: does the engine know who you are?
4. Organization JSON-LD on the homepage
Name, URL, logo, contact email, phone, address (at least city and country), and sameAs links to LinkedIn and any other profiles. This single block is how the engine connects your site to the entity it already has in its index.
5. The same facts everywhere
Company name, city, founder, services. Check your site, LinkedIn page, Google Business Profile and the top three directories you are listed in. A business described as "Dubai" on one and "UAE" on another, or with two different founder names, gets treated as two uncertain entities rather than one confident one.
6. A real person on the About page
Engines are increasingly cautious about recommending anonymous companies. A founder name, short bio and LinkedIn link, with Person markup, is a trust signal that costs nothing.
7. An llms.txt at the root
A plain-text summary of what you do, your key pages and the facts you want repeated. Not every engine reads it yet, but the ones that do use it directly, and it forces you to write your own one-paragraph description, which you will reuse everywhere else.
Content: is there anything worth quoting?
8. Every service has a one-sentence definition
Under each service heading, the first sentence should answer "what is this and who is it for" in plain language. Read it aloud. If it would sound natural as the first line of an AI answer, it is right. If it sounds like a slogan, rewrite it.
9. Numbers on every project page
Dataset size, accuracy, number of cameras, number of users, years of data, languages supported. Engines quote specifics because specifics make their answer look grounded. A project page with zero numbers is a project page that will not be cited.
10. An FAQ with FAQPage markup
Five to ten questions phrased the way a buyer would ask an assistant, each with a two- or three-sentence answer. This is the closest thing to a direct feed into an AI response that exists.
Authority: why you over the other five results?
11. Published, dated content under a named author
A blog or insights section with Article markup, a real author and a publish date. It does two things: it gives the engine more retrievable passages on the topics you want to own, and it signals that the site is maintained. Two good articles a month beats ten thin ones.
12. Mentions you don't control
Client case studies on the client's site, directory listings, public GitHub repositories with READMEs, talks, guest posts. The engine weighs a third-party page describing you far above your own description of yourself. This is the slowest item on the list and the one with the highest ceiling.
How to know it's working
Write down 15 to 20 real buyer questions. Ask ChatGPT, Perplexity, Gemini and Google (for AI Overviews) each of them once a month, in a fresh session, and record three things: are you mentioned, are you linked, and who else is named. Keep the spreadsheet. Movement usually shows within four to eight weeks of fixing items 1 through 10; item 12 keeps paying out for years.
If you want a shortcut for the site-side checks, AIGEO scores a site 0 to 100 against most of this list automatically, and our AI Visibility service covers the rest.