Your customers
ask ChatGPT.
Does it say your name?
Buyers ask ChatGPT, Claude, Gemini and Perplexity for a shortlist before they open Google. We measure how often those answers mention you, then do the work that gets you cited.
Based on recent sources, these Dutch agencies are frequently recommended:
- 1
Agency A
Cited from an industry directory and two client reviews.
- 2
Agency B
Cited from a comparison article and a podcast transcript.
- 3
Agency C
Cited from its own case-study pages.
Your company is not mentioned. The buyer never sees your name.
Illustrative. Your baseline scan shows the real picture.
People still search.
They just get an answer instead of a list.
A buyer used to get ten links and pick one. Now they get a paragraph with three company names in it. If yours is not one of the three, there is no click to lose. You simply never hear about it, and it does not show up in your analytics either.
25%
drop in traditional search engine volume by 2026 as buyers switch to AI chatbots, Gartner predicted.
Gartner, Feb 20248%
of Google searches that show an AI summary end in a click on a result. Without the summary: 15%.
Pew Research Center, Jul 20251%
of visits with an AI summary clicked a source cited inside it. Almost nobody clicks through. The name in the answer is what the buyer remembers.
Pew Research Center, Jul 2025282M
Google AI Overview queries a month in one tracking index, against roughly 15M each for ChatGPT, Gemini and Perplexity. Most people already see AI answers without opening a chatbot.
Ahrefs Brand Radar, 2026Four things decide
whether you get mentioned.
None of this is secret. It is work, and in most Dutch categories nobody has done it yet.
They trust what others say about you
Models put more weight on independent sources than on your own site. Directories, reviews, trade press, comparison articles, forum threads. If nobody else describes you, the model has nothing to repeat.
They need to know what you are
The model has to connect your name to a category, a location and a type of customer. That takes the same name everywhere, a plain description of what you do and for whom, and structured data that says the same thing in machine-readable form.
They quote pages that answer the question
Pages that answer a concrete question, with real numbers and clear structure, get cited. Brand copy about your mission does not. In practice comparison pages, pricing pages and FAQs do most of the work.
Each model reads a different web
ChatGPT, Claude, Gemini and Perplexity each mix live search and training data in their own way. Showing up in one says little about the other three, so we measure all four.
What you get.
Five things, all of them checkable.
Baseline scan
We write the prompts your buyers actually type, in Dutch and English, and run them across ChatGPT, Claude, Gemini and Perplexity. You get your share of voice, who gets named instead, and which sources feed every answer.
Citation gap analysis
A list of the pages and publications the models trust in your category, ranked by how often they are cited, and where you are missing from them.
Entity and site work
About and service pages that state plainly what you do, structured data, comparison and FAQ pages that answer the question directly, an llms.txt file, and crawler access for the AI systems that index the web.
Third-party presence
The directories, review platforms, industry publications and communities that show up in the citations. Real listings and real mentions. No paid link schemes, no fake reviews.
Monthly re-measurement
Same prompts, same models, one report. Mentions, position, cited sources, movement against the baseline. When a number does not move we say why and what changes next.
First numbers within a week.
Moving them takes a quarter.
Week 1
Scan
Prompt set, four models, baseline numbers. You see exactly where you stand.
Week 2
Plan
Gap analysis and a fixed-scope plan. If the gap is small we tell you and stop here.
Weeks 3 to 10
Build
Site work, structured data, third-party presence. Shipped in batches you can review.
Monthly
Measure
Re-run, report, adjust. Ongoing as long as the numbers earn it.
Four rules we hold ourselves to.
Measure first, sell second
The scan is fixed-scope and comes before any proposal. If it turns out the models already name you, or nobody asks about your category, we tell you and there is nothing more to buy.
You own the data
The prompt set, the raw answers, the source lists and the reports are yours, delivered as plain files. There is no dashboard subscription to get stuck in.
Dutch and English
Buyers in the Netherlands ask in both languages, and the answers differ. We measure and write in both. Most tooling on the market only looks at the English web.
Privacy-first, as always
No tracking pixels on your site, no scraped personal data, nothing sent to a model you have not approved. Same rules as every other system we build.
Questions we get on the first call.
Partly. Technical SEO still matters because several models pull from search indexes. The difference is what happens after. A search engine lists ten links and lets the buyer choose. A model reads a handful of sources it trusts and writes one answer. So the question shifts from "do we rank for this keyword" to "are we in the sources the model reads, and do those sources describe us well". Third-party pages count for more than your own.
Models that search live (Perplexity, ChatGPT with browsing, Gemini) can reflect new pages and mentions within weeks. Anything that depends on training data moves slower and less predictably. We report monthly and are honest about which is which.
No. We run the measurement. If you later want a dashboard, we will recommend one, but the prompt set and the data stay yours either way.
Usually yes, and often more than for a big one. In most Dutch categories nobody is doing this yet. A clear description of what you do plus a handful of good third-party mentions can already change the answer. The scan tells us within a week whether that holds for your category.
The baseline scan is a fixed price. Ongoing work is quoted after the scan and scoped to the gaps we actually find. Ask for the current rate on the call.
No, and nobody can. Models change weekly. What we guarantee is a real baseline, a plan built on what the models cite today, and a monthly number you can hold us to.
Find out what the models
say about you today.
Thirty minutes. We look at your category together and decide whether a baseline scan is worth doing.
Also building automation? See what else we do.