Get recommended by ChatGPT: how AI assistants come to name your business

How do AI assistants decide which business to name?

When someone asks "Which electrician in Berlin-Neukölln is reliable?" or "Where can I get a good breakfast in Kreuzberg?", the assistant builds its answer from two sources: what the language model has learned, and a live web search. ChatGPT, Gemini, Perplexity and Google AI Overviews weight these sources differently, but the principle is the same.

The businesses that get named are the ones that clearly fit the question and whose details can be confirmed in several places. An assistant does not want to give a wrong recommendation. The more clearly your business is described and the more often the same facts appear in different places, the more confidently it can name you.

Generative Engine Optimization, or GEO, covers everything that supports this. It builds on classic search engine optimisation and does not replace it.

Why does consistent business data matter so much?

Name, address and phone number, known as NAP, are the foundation. To an AI system, "Example GmbH, Hauptstr. 5" and "Example Berlin, Hauptstraße 5" are initially two different things. Only when the details are identical everywhere does the system merge them into one entity.

So check every place your business appears and write the business name, address, phone number, opening hours and services exactly the same way:

• →Website: home page, contact page, imprint, footer.
• →Google Business Profile and Apple Maps.
• →Business directories, chambers and trade associations.
• →Social profiles such as Instagram, Facebook and LinkedIn.
• →Review platforms and booking portals.

What do structured data and llms.txt do?

Structured data based on schema.org describes your business in machine-readable form, directly in your website's source code. For local businesses the key building blocks are Organization or LocalBusiness with address, phone number, opening hours and service area. Add Service for each offering and FAQPage for common questions. The sameAs field links your website to your Google profile, Instagram and other official profiles.

The llms.txt file sits in the root of your website, at yourdomain.com/llms.txt. In plain text it summarises who you are, what you offer, where you operate and how to reach you. The standard is still young and not every system uses it the same way, but the effort is small and the file forces you to write a clear self-description you can reuse elsewhere.

How do you write service pages that get quoted?

AI assistants like to pick up short, unambiguous statements. So open every service page with a one or two sentence definition that makes sense without any further context. A pattern: "[Business name] is a painting company in Berlin-Tempelhof offering interior painting, facade painting and wallpapering for private and commercial clients."

After that, headings phrased as questions, answered directly by the first paragraph, help people and machines find the right passage quickly. Other things that make pages quotable:

• →One page per service instead of a catch-all page.
• →Specific details on service area, process and audience.
• →Tables and lists for facts that can be compared.
• →A visible date of the last update.
• →An FAQ with the questions customers actually ask you.

What role do reviews and mentions on other sites play?

What you write about yourself is only one source for an assistant. Confirmation comes from outside: reviews on Google and review platforms, listings in business directories, coverage in local media, mentions by partners, suppliers or associations.

Actively ask happy customers for a review and reply to every one, including critical ones. Reviews that mention the service and the location help most, because they confirm your self-description in other people's words. Never buy reviews: it breaks the platforms' rules and can put your profile at risk.

Are AI crawlers allowed to read your website?

Many websites lock out AI crawlers without meaning to, for example through an overly strict robots.txt or hosting defaults. An assistant running a web search then cannot fetch your pages and falls back on other sources.

Check that your robots.txt allows these crawlers: GPTBot, ChatGPT-User and OAI-SearchBot (OpenAI), PerplexityBot, ClaudeBot, Google-Extended, Bingbot and Applebot-Extended. Also make sure important content is in the HTML as text, not only in images or loaded later by scripts.

Clieni Media checks these points first in every GEO project, because everything else depends on them. If you would like to see where your own website stands, try the free website check at Website check.