AI Search
Getting named and cited at the moment your customer asks their question.
What AI Search is
AI Search is making your brand visible in the answers of AI search engines: ChatGPT, Perplexity, Claude and Google AI Overviews. Not as a mention tucked away at the bottom, but as one of the sources the answer is built from. The trade also calls it GEO, generative engine optimisation, or AI search optimisation; the work behind it is the same.
The difference from classic search is where the choice gets made. A search engine shows ten results and lets the user choose. An AI search engine writes a single answer and picks a handful of sources itself.
That playing field is smaller, and whoever is on it takes more.
Being there when the need arises
Someone weighing up their options increasingly puts the question to an AI search engine, often before they have any supplier in mind. That is the moment the shortlist forms that you will later be competing against.
If you are named there, you reach prospects you did not know and who did not know you. That is exactly the reach your network cannot give you: people who have a need today, are actively searching, and have no preference yet.
If you are not there, you never notice. No lost enquiry comes in, and nobody calls to tell you that you were not mentioned.
It is compounded by the growing share of questions answered without anyone clicking through. We see it in the measurements we run for our clients: since AI answers started appearing above the regular results, brands get named more often and clicked less often. More people see something about your market, fewer people end up on your site.
The deal is lost before the contact ever happens.
What it delivers
Done well, you get named as the solution to the questions your ideal customer asks. That produces a different kind of contact from an ad or a cold approach: someone who is actively searching, has a concrete need, and already sees you as a serious option before the conversation starts.
Research shows the same picture. In 2025 Semrush found that visitors arriving via an AI search engine are roughly 4.4 times more valuable than visitors from regular organic results, measured by conversion. The explanation is straightforward: by the time someone clicks through, the model has already summarised and compared their options. They are not arriving to look around; they arrive because they already know what they want.
We see the same with our clients. You come in at the moment the need exists, so the first half of the conversation has already been had. Add to that the way an answer from an AI search engine lands differently from a search result: it is presented as the answer, not as a list of options, and that gives a recommendation more weight.
There is less explaining who you are and what you do. The lead cycle gets shorter.
How AI search engines choose their sources
An AI search engine answers a question in three steps: it works out what is meant, retrieves candidate sources, and writes an answer from the sources that answer the question most clearly. Sources drop out at every step.
It also weighs your brand as a whole: a name, a specialism and a reputation that read the same in several places. The clearer that picture, the smaller the chance you are left out through doubt or confused with someone else.
For AI search engines, authority is mostly consistency.
Every platform chooses differently
The same brand scores wildly differently from one platform to the next. Optimising for a single platform is therefore always half a job.
ChatGPT
Combines what the model already knows with live search, leans on the Bing index for that, and cites Wikipedia strikingly often.
Perplexity
Looks up almost every question live, works semantically, and leans heavily towards user-generated sources such as forums and reviews.
Google AI Overviews
Builds on the regular Google index, so existing rankings count heavily, and pays less attention to the domain than the others.
Copilot
Leans heavily on authoritative publications. If you are not in those, it is hard to get in.
The models we have insight into
We do not measure a single model. These are the AI search engines and models for which we see, per question, whether you get named, what is said about you and which sources carry the answer.
- ChatGPT
- Google AI Overviews
- Google AI Mode
- Gemini
- Perplexity
- Claude
- Copilot
- Grok
- DeepSeek
- Mistral
- Llama
- Alexa
- Kimi
- GLM
- GPT OSS
- Claude Code
- OpenCode
- Inkling
What we do
Measure what is happening now
Per model, against your competitors, with the questions your customers actually ask. Repeated over a longer period, so you end up with percentages rather than a snapshot.
Get the technology right
Let AI crawlers in, add structured data, publish an llms.txt, and make sure your pages are readable for systems that read far less patiently than a search engine. Most AI crawlers cannot handle JavaScript: if your content only appears once a script has loaded, it does not exist for those systems.
Show up on the questions that matter
We make you visible to your ideal customer on the questions that matter to them. Which ones those are follows from the measurement: it shows you where you are missing today and where a competitor is giving the answer.
Align the brand picture
The same name, description and facts on your site, your profiles and the sources that write about you. Remove contradictions, because they cost you trust with the model.
Build authority beyond your site
Get named on the sources the models weigh heavily. Every extra mention shifts the weight in your direction.
AI Search does not replace SEO
SEO and AI Search draw partly on the same sources. A healthy technical base and strong content help both. Drop SEO in favour of AI Search and you undermine exactly the sources the models lean on.
What differs is the yardstick and part of the writing. SEO measures rankings; AI Search measures your share of answers. A page that ranks well does not have to answer the question in its first paragraph. A page that gets cited does.
Part of a bigger whole
AI Search is one of the methods we use. What exactly needs to happen, and in which order, follows from the measurement and the strategy. Without that picture, optimising is guesswork.
If a Web Intelligence programme is running, we measure what every change delivered, per model and against your competitors.
Frequently asked questions
Is GEO the same as AI Search?
Yes. GEO, LLM SEO, AEO and AI search optimisation all describe the same work: making sure language models name your brand and cite your pages, so that you can be found in ChatGPT and the other AI search engines. The terms vary by field and by country; the work behind them does not.
Which AI search engines matter most?
For most Dutch and Belgian markets: ChatGPT for its reach, Google AI Overviews because they sit above the regular results, and Perplexity and Claude as the fastest-growing sources of click-throughs. The mix differs per audience, so we measure per model before setting priorities.
Do I have to rebuild my website?
Usually not. Targeted repair work is often enough: letting crawlers in, adding structured data, rewriting headings and opening paragraphs. A rebuild only comes into view when the technical base is so weak that every improvement founders on it, and the measurement shows that well in advance.
How long before something moves?
That depends on where the bottleneck is. If the picture comes from sources that are fetched live, the effect is visible within weeks. If it sits in the model itself, it moves over quarters. The measurement shows per question which case you are in.
Does this work for small companies and niches too?
Especially there. Models weigh the quality of an answer more heavily than the size of a company. And in a niche the field of questions is smaller, so it is quicker to claim.
What is an llms.txt?
A text file on your domain that tells AI systems in plain language what you offer and which pages matter. It is new and no guarantee, but it costs little and makes your site more predictably readable.
Not seeing your question?
Get in touch, no obligation. We are happy to talk through what this would mean in your situation, even if you are not sure yet whether it is something you need.
Get in touch














