GEO and AI Search Visibility

AI search visibility: how assistants choose what to mention

When someone asks an assistant to recommend a clinic or a contractor, the answer is built from pages the assistant was able to read. It tends to use sources with four things in common.

• They are crawlable. The assistant's crawler is allowed in.
• They are clearly structured. Headings, questions and answers are easy to find.
• They are consistent. The site always says the same thing about who the business is and what it does.
• They are confirmed by other sources. Profiles, directories and other sites repeat the same details.

None of this is secret. It is careful housekeeping, done properly.

What generative engine optimisation is

Generative engine optimisation, or GEO, is the work of making a business easy for AI assistants to understand and quote. Classic SEO aims at a position in a list of results. GEO aims at being named inside an answer.

The two are not rivals. A page that cannot be crawled or indexed has little chance of being quoted anywhere. GEO depends on classic SEO and adds a layer on top: who you are, stated the same way everywhere, and content written to answer a question directly.

That is why this work sits inside a monthly plan, next to the rest of my SEO services.

The work itself

Entity consistency

The same name, address and services on the website, the Google Business Profile and every listing.

Schema markup

Structured data that tells machines what the business is, where it is and what it offers.

Answer-first content

Pages built around clear questions, with the answer in the first sentence.

AI crawler access

I check that robots.txt does not block crawlers such as GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot.

Arabic and English coverage

The business is described in both languages, with the Arabic written natively.

Where the problem is technical, I start with a technical SEO audit.

How it is measured

You cannot look up a ranking inside ChatGPT. So I measure it the only honest way: by asking.

At the start we agree a fixed set of prompts, the questions a customer would really ask about your service. Every month I run the same set and record two things: whether the business is mentioned, and whether the website is cited as a source.

The prompts stay the same, so one month can be compared with the next. The results go into the monthly report, next to rankings, calls and forms.

What nobody can promise

No one can guarantee a mention in an AI answer. The assistants decide, and their answers change.

Two things are often sold as shortcuts. Schema markup helps machines read a page, but it does not guarantee a citation. An llms.txt file is easy to add, and it does not guarantee one either.

What can be controlled is smaller and more useful: whether the site is readable, whether it is clear about what the business does, and whether it is consistent everywhere it appears. That is the work I do, and that is what I report on.