A growing share of buyers asks an AI tool first and only then opens a search engine. The answer they get contains three to five company names. Optimising for those answers is called AEO — and it rests on different signals than classic SEO.
How it differs from SEO
Classic SEO competes for a place on a list of links; the user then chooses. An AI answer makes the choice itself and names a limited number of companies. So keyword breadth matters less, and three things matter more: clarity about what you do, confirmation of that data across several sources, and the machine readability of your site.
Start by checking where you stand
Run the same prompt in ChatGPT, Gemini and Perplexity: “I am looking for a proven supplier of [service] in [region]. Which companies would you recommend and why?” Record which companies are named, which sources the models cite and whether the data about you is correct. The detailed procedure is in the guide how to check what ChatGPT says about your company.
Write pages that answer questions
Models prefer content a statement can easily be extracted from. In practice: a heading that is a question or a clear claim, the answer in the first two sentences, then the detail. Instead of an “Our services” page with five generic paragraphs, a page titled “Laser cutting of sheet metal up to 20 mm” with materials, tolerances, lead times and delivery area is far more useful.
Concrete data — dimensions, capacities, lead times, service areas, certificates — is what a model can quote. Phrases like “years of experience and top quality” carry no information and are worthless in an answer.
Mark your data up for machines
Structured data (schema.org) is the format in which a page states what it is: an organisation, a service, a frequently asked question, a product. The minimum set for any company is organisation markup with name, address and contact, service markup, and FAQ markup. Alongside that come the basics that often get missed: a correct page title, a description, unique URLs, and a page that loads without blocking.
Confirmation outside your own site
A model treats a fact that appears only on your own site as a claim, and a fact repeated across several independent places as established. That is why your Google business profile, chamber and trade directories, relevant B2B portals, mentions in trade media and customer reviews all matter. More detail in where AI looks for confirmation.
Consistency matters more than volume here: the same name, the same address, the same description of your activity everywhere.
The order we recommend
1. A snapshot across three tools and a list of incorrect or missing data.
2. One page per service or product, with concrete technical data.
3. Structured data and FAQs on those pages.
4. Aligning external profiles and directories.
5. A re-check after six to eight weeks, because data enters the models with a delay.
The goal is not to outsmart the model, but to give it verifiable data it can repeat without risk.
Related: why your website does not appear on Google · the AI visibility service