GEO audit: what we check, point by point

A GEO audit is not a score out of a hundred. It is a list of observable facts, ordered by priority, that says where you are missing from generated answers and why.

What AI says about you today. We start by questioning the engines the way a client would: same questions, in French and Dutch, several phrasings. We note whether you are cited, what is said about you, and the factual errors.

Pages robots can read. Useful text must be in the code, not only in an image or a script. One title per page, a clean heading hierarchy, direct sentences.

llms.txt. This is a text file at the root of the site listing your main pages for AI agents. The format is not recognised by every engine, but it is simple to add and costs nothing. We treat it as a complement, never as a guarantee.

Structured data. We check LocalBusiness, Service, FAQPage, opening hours, the area served and sameAs links. We validate the markup and fix errors, because wrong markup is worse than no markup.

Profiles and directories. Google, maps, professional directories, federations, chambers, booking platforms: we list every mention, align name, address and phone, and remove obsolete duplicates.

Reviews. We look at the number, freshness, rating distribution, replies and platforms. An engine that summarises your business relies heavily on what clients write about you.

Content that answers real questions. We compare existing pages with the questions actually asked. The gap is often stark: many pages talk about the company, few answer “is it reimbursed?” or “what happens during the first session?”.

Third-party mentions. We look for who talks about you: press, partners, associations, blogs, directories. We note the quality and consistency of those mentions, because they weigh on an engine's decision.

Bilingual consistency. If you are based in Charleroi and also target Flanders or Brussels, we check that both versions say the same thing, with the right words in each language.

Before/after measurement. We freeze a set of questions and queries, record it before the work, then at regular intervals. We also track brand searches, clicks and incoming enquiries. We set a reference date, then compare at three-month intervals. Without a starting point, any improvement becomes an impression, and decisions are made on intuition rather than facts.

What the engines cite instead of you. We also list the businesses that appear in answers to the questions you should be winning. Sometimes they are legitimate competitors; sometimes they are directories, platforms or media that simply wrote the right page, in the right place, with the right words. That list tells you who you are up against, and on which ground you need to be present.

The deliverable. A useful audit fits on a short list: what is broken, what is missing, what can be fixed this week and what needs a project. An overall score is useless if it does not say where to start. We add an effort estimate, because a priority without a stated effort is not a priority, it is a wish.

A concrete reference: Centre Vital shows a rating of 4.8/5 from 49 reviews, 742 clicks for 43,823 impressions in three months and an average position between 4 and 5 on its main queries. These are visible traces, therefore citable. An audit looks for exactly that kind of material.

What an audit does not do. It does not guarantee a future citation and does not replace sustained work on quality and reputation. It makes the problem visible, which is already a lot.

Curious what the engines say about you? Book a free 30-minute discovery call: we run the first checks live and show you the three priorities that would make the biggest difference.