AI visibility · audit
I run a fixed set of 40 buying questions through ChatGPT, Perplexity, Google AI Overviews and Copilot, keep the raw answers, and name every company the models put in front of you. You get a report and a prioritised fix list, not a ranking promise.
4
engines in one run
40
questions with raw answers
10
working days to the report
A sample run, not a client. Your numbers will differ, and the measurement changes nothing on its own: the fix list under it does.
Four things I see in almost every business before anything is measured.
Where an AI Overview appears, click-through drops by roughly 34.5% (Ahrefs, 300,000 keywords), and overviews now show on around 92% of informational queries. Position three stays position three, it just stops earning clicks. Analytics shows the dip without the reason.
You type your category into ChatGPT once, see a competitor, and draw a conclusion. The same prompt an hour later returns a different list. Without repeats and saved answers you have an impression, not a baseline you can compare against next quarter.
No AI crawler executes JavaScript. GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot fetch the HTML and stop there, as of June 2026. Prices behind a tab, FAQs in an accordion and reviews loaded after hydration simply do not exist for them.
In a study of 278 B2B SaaS buying prompts (July 2026), 66% of the times ChatGPT recommended a brand it cited that brand's own website nowhere in the answer. The recommendation came out of other people's roundups, mentions and threads. On-site work alone cannot buy it.
One file plus an appendix with the raw answers from all four engines.
We build 40 prompts your buyer actually types: the category, the city, alternatives to a named competitor, "who can build me X". The set is frozen so the next run is comparable to this one.
Every answer from all four engines is saved with its date and the model version used. You do not have to take my word for anything, you can read what the model wrote.
I list every company that surfaces in answers to your questions and show what put it there: a roundup, a directory, a forum thread, a trade-press mention.
I check what survives of your pages in plain HTML with JavaScript off, and how bots are treated in robots.txt and at the firewall. Prices, FAQs and reviews that are invisible get listed separately, URL by URL.
Best-of roundups, directories (Clutch, GoodFirms, DesignRush, vertical listings), review platforms, Reddit threads, and whether your company name, description and registration details match across every profile.
Each item carries an effort estimate, who should do it, and whether it sits on your site or off it. A separate block lists what I recommend against, with the reason.
Start on a Monday, report and call two weeks later.
You send 10-15 phrases buyers use to find you and a list of competitors. I turn that into 40 questions and send them for sign-off. Nothing runs before you approve the set.
Every question goes three times into ChatGPT, Perplexity, Google AI Overviews and Copilot, from a clean session each time. Answers are stored in full, together with the sources each engine listed.
I check what is left of your pages without JavaScript, how bot access looks in robots.txt and at the firewall, and whether your company details agree across the site, your profiles and the directories.
I find the roundups, directories and threads where competitors are named and you are not. For each one I write down who decides on inclusion and what they ask for.
You get the file, the appendix of raw answers and 60 minutes on a call. You leave with a task list you can hand to anyone, not only to me.
Packages
One measurement and the report. You or your team apply the fixes.
Typical fit: a B2B service company, one site, one market.
Everything above, and I do the technical work on your site myself.
Next.js, WordPress, Shopify. A bespoke CMS is quoted separately.
The same question set every month and a report against the last one.
Worth doing after an audit, not instead of one.
I do not sell positions or guaranteed citations, because nobody controls them. I sell the measurement and the fix list.
Four cases where I talk you out of it instead of selling.
If the real problem is the site itself rather than your visibility in AI answers, start here: Services
Describe the task in two sentences. You get a fixed price, a timeline and a plan. If it isn't a fit for us, we say so straight away and point you elsewhere.
No, and nobody can. Models assemble answers from sources neither you nor I control, and the answer shifts between runs and between model versions. What I sell is the measurement, the named list of who gets recommended instead, and an ordered fix list. The effect is checked by re-running the same question set later.
Because a single answer means nothing. The same prompt asked three times can return three different lists of companies, so what counts is how often you appear across repeats, not whether you appeared once. That is why the report gives you a fraction like 3/40 rather than the word "visible".
Because that is where recommendations are made. In a study of 278 B2B SaaS buying prompts (July 2026), 66% of ChatGPT's brand recommendations carried no citation of the brand's own site, and the strongest correlate of being recommended was appearing in third-party roundups, followed by branded web mentions and a Reddit presence. On-site work removes technical blockers, but it does not buy the recommendation.
No. Ahrefs looked at 137,000 sites and found that 97% of llms.txt files received zero traffic in May 2026, and Google has said plainly that it does not support the format and has no plans to. No major model provider has committed to it either. If you already have the file, leave it, but it does not count as visibility work.
Yes, and the problem is a specific one. AI crawlers fetch the HTML and do not execute JavaScript, so prices, FAQs and reviews painted in the browser do not exist for them. The audit gives you the list of URLs and blocks missing from the page source, and in the package with fixes I move them to server rendering.
The audit is €900 one-off, the audit with the technical fixes applied is €1,600, and the monthly re-run is €350 per month. The difference is simple: the first gives you the report and you decide what to do with it, the second means I do the on-site work, the third tracks whether the picture moves. The number is fixed before we start.
The honest answer is that nobody knows in advance, which is exactly why we re-measure. Technical fixes take effect from the crawler's next pass, tidy company data tends to show up in brand recognition after a month or two, and getting into other people's roundups depends on their editors rather than on us. So I re-run the same questions and show the difference instead of quoting a date.
What clients say
A few words from founders and teams we've shipped for.
Launched our store in 11 days — multilingual, fast, and already on page one for three of our main keywords.
The Telegram bot handles the first conversation with every lead now. My sales rep only steps in when the deal is hot.
Leads land in our CRM with every field already filled. No more copy-pasting from chats — the pipeline moves itself.
Shipped to the App Store and Google Play in three weeks. The handover docs let my team take it from there without me.
Set up our Google and Meta ads from zero. Cost per lead dropped by about a third in the first month.
He rebuilt the site and six weeks later ChatGPT was recommending us by name. Didn't know that was even possible.
Free 30-minute call. No sales pitch - just a real conversation about what you need and whether we're the right studio to build it.
No sales pitch — just a clear read on your scope.