SEO & AI Search

How to Check Whether Your Website Is Ready for AI Search

A half-hour checklist: AI crawler access, answers in the text, structured data, content worth citing, speed and indexing. No promises about getting you into ChatGPT.

When someone asks ChatGPT, Perplexity or Google's AI Overview who they should trust with a website redesign in their town, the answer is assembled from pages the model was able to read and understand. Asking whether your site is among them is a reasonable question. Less reasonable are promises that someone will "get you into ChatGPT". Nobody can guarantee that. What you can do is check whether your site even meets the conditions to be considered.

Here is the list I use myself. Most of it you can do in half an hour without tools; for a few points an automated check helps.

Short answer

A website is ready for AI search when AI crawlers are allowed to read it, when the answers to readers' questions are written plainly in the text (not only in images or scripts), when structured data says who you are and what you offer, and when the content is concrete enough to be worth citing. All of that is the same thing that's good for classic Google; AI search just punishes vagueness harder.

Are AI crawlers allowed in

The first step is robots.txt. Check whether it blocks the crawlers AI search engines use. There's a distinction many sites miss: OpenAI uses a separate crawler for search (OAI-SearchBot) and a separate one for model training (GPTBot); Anthropic likewise separates Claude-SearchBot from ClaudeBot. If you block everything with "bot" in its name, you've removed yourself from AI answers too. On this site I allow the search crawlers and decide separately about the training ones; I wrote about that in SEO Is No Longer Just Google.

Also check whether your server or bot protection (for example CDN settings) returns an error to these crawlers. You can't see that from a browser; you see it in server logs or with a tool that visits the page as a crawler would.

Is the answer in the text

AI models cite text. If your main service is described in an image, in a video, or in text that loads only after a click, it doesn't exist for the model. I check by opening the page without images and JavaScript, or simply by reading the source: is there a sentence that says what you offer, to whom and where?

The same goes for questions. If customers ask how long it takes, how much it costs and whether you work on Saturdays, the answer should be on the page as an answer, not as "contact us". A subheading phrased as a question and a first paragraph that answers it directly is the form models use most easily. It's not a trick; it's just clear writing.

Does the site say who you are

Structured data (JSON-LD following schema.org) is how you tell a machine that this is a business with a name, address, phone number and opening hours, that this page is an article by this author, and that this list of questions is an FAQ. For a local business, LocalBusiness with a real address and opening hours is the most useful piece. Check it with Google's Rich Results Test or the schema.org validator.

While you're there, check the basics: does every page have its own title and meta description, is your about page linked to your public profiles, is the author of your articles a real, named person. Models look for signs that a concrete person stands behind the content.

Is the content worth citing

This is the hardest point and no tool solves it. A page of generic sentences ("we offer quality services at affordable prices") won't be cited, because it says nothing a model could use. A page that explains concretely how you work, what's included and what isn't, has something to offer. When I write for this site I ask myself: if someone quoted this paragraph, would the quote be useful on its own?

An llms.txt file helps here, but it isn't the solution. It's a simple list of important pages with a short description for models that read it. I have one; I don't believe it changes anything by itself if the content behind it isn't good.

Does the site work fast and on a phone

AI search engines often use the same index as classic search engines, so the technical basics still apply: speed, mobile layout, working links, no duplicate titles, no stray noindex tags. Google PageSpeed tells you where you stand; Search Console tells you which pages are in the index at all.

How I check it with a tool

A manual review is good for understanding, but for a whole site it takes time. For a quick picture I use Sitequiry, a tool we built at GE-KO for exactly this: for one URL it checks technical SEO, speed via Google PageSpeed, and readiness for AI answers and citations (crawler access, structured data, clarity of content) for free, and returns a report with scores and concrete next steps. It's not the only tool of its kind, and no tool can see whether your content is genuinely useful; it does show the technical gaps you'd otherwise miss.

The half-hour checklist

  1. Open /robots.txt and make sure AI search crawlers aren't blocked.
  2. Read the source of your home page: is it clear in the text what, for whom and where.
  3. Validate your structured data.
  4. Open the three pages you'd most like to be cited for and check whether the first paragraph under each subheading answers the question.
  5. Run PageSpeed for mobile.
  6. In Search Console, check which pages are indexed and whether there are crawl errors.

If the gaps you find are technical, they can be fixed in a few hours. At GE-KO that's part of our AI search optimisation work. If the gaps are in the content, they'll have to be written, and that's work nobody can do for you, because only you know how you work.

Photo of the author

I build web solutions and digital products. At GE-KO I work on web design, development and SEO; with Bizko I look for simpler ways through everyday business. I am interested in how technology saves people work, and in how AI is changing the web.

One step further.