If someone asks ChatGPT, Gemini or Google's AI Overviews who solves their problem and you're not the answer, you've lost the deal before your site ever comes up. We track what your buyers actually ask, check how AI describes you, read what AI agents fetch from your site, and build the presence that gets you cited, tied back to the leads it produces.
Noa NotesB2B SaaS · digital health platformNoa Notes, Jameda's AI assistant for clinical documentation, is part of Docplanner's platform for doctors and patients.
First 6 monthsB2B or B2C, your buyers now ask an AI about your category before they ask you.
Whether it's a buying committee comparing vendors or a shopper comparing products, everyone checks ChatGPT or Gemini first, before they ever reach you.
Google's AI Overviews and AI Mode sit inside the results people use every day. That's the default now, not an option to opt into.
A consumer-facing bank or insurer meets the same AI-driven research habits as any B2C brand, just at far bigger scale.
It rarely looks like a dramatic drop. You lose the recommendation moment first, a discovery-layer gap that isn't a line item you can defer.
A tool gives you a dashboard. We set the tracking up properly, read what it says, turn it into a list of work, and then do the work.
Monitoring
There is no reliable search volume for prompts, so we build your prompt set from what buyers actually say: sales calls, reviews, support tickets and Search Console. We expand it into the fan-out queries an engine runs on its own and track it every day in the engines your buyers use, so the numbers describe your market rather than a keyword tool's guess at it.
Built from what your buyers already say
Tracked daily in the engines your buyers use
Read as numbers you can report upward
Brand perception
For a large brand, being mentioned is rarely the problem. The problem is an answer that quotes last year's pricing, names a product you retired, or tells a buyer you only serve small companies. We read the answers for facts, tone and the objections they raise, trace each wrong statement back to the page it came from, and get that page corrected, on your own site or wherever it lives.
Facts checked against your own sources
Perception compared with your competitors
When it is wrong we fix the source
Agent analytics
Prompt tracking shows the answers. Your server logs show the other side: which AI bots visit, which pages they fetch and which ones they never reach. A training crawler, a search indexer and an assistant opening a page because someone just asked a question are three different signals, so we keep them apart. What comes out is a short list: the content AI keeps coming back to and deserves more of, the topics it ignores, and the errors or blocks that keep a page from being cited at all.
Three kinds of visit read separately
What we take from it and put on your roadmap
Execution
Every finding above becomes an item on your roadmap, and our team ships it: technical fixes so AI crawlers can reach and parse your pages, content structured so an engine can lift the answer, and work off your site wherever the citations in your category actually come from. Which platforms matter is a finding from your data, never a default.
On your site the surface you control
Off your site where your category gets cited
We track your AI visibility score just as closely as the leads it produces, since looking at only one tells half the story.
Niklas Buschner hosts Masters of Search, more than 45 episodes with operators in AI Search and SEO. What they say feeds directly into how our team runs accounts.
From AEO versus GEO to what happens when an engine gets your facts wrong.
Both, honestly. We think of them as two gears turning together, not two separate services. SEO fundamentals, being the best answer on the open web, are the base gear. Turn it, and it drives a second one: whether ChatGPT, Microsoft Copilot, Gemini, and Google's AI Overviews cite you for the question you'd want to own. On top of that sit a handful of initiatives built specifically to turn that AI-visibility gear faster: a real, validated prompt set, tracked to see which sources actually get cited, then work placed exactly where those citations land, whether that's your own site, a YouTube video, or a review platform. AI Search is grounded in fundamentals we've run since 2013, technical health, genuinely being the best answer on a topic. But it also opens surfaces SEO never had to touch on its own, and comes with an attribution problem SEO never had either, since last-click reporting simply doesn't see an AI recommendation. Anyone who tells you it's identical to SEO, or a completely separate discipline, is oversimplifying to sell you something.
Yes, and it's a focus we're actively growing. Large organizations research differently: several people across a buying committee independently ask the same category question, not just one champion, so visibility has to hold up across all of them, not a single journey. We work alongside whatever internal team or agency of record you already have, inside your existing tools and approval process, rather than asking you to route everything through us. The one thing that doesn't change with company size: we still commit to a concrete goal and report against it the same way.
A senior team, the whole way through. There's no senior-in-the-pitch, junior-in-delivery handoff here: a senior strategist owns your account from day one, works inside your own Slack or Teams, and stays hands-on for as long as we work together. The same senior person builds your prompt set and reads the results, not a junior translating it back to you afterward.
It's part of the same Growth Partnership retainer, starting at €7k a month. We don't think SEO and AI Search should be split into separate line items, since the work genuinely overlaps. For something narrower, a YouTube optimization push, say, we also run fixed-scope projects instead, priced separately and over a shorter, one-off timeline rather than an ongoing retainer.
Not much in practice. Answer Engine Optimization and Generative Engine Optimization are two names for the same shift: getting cited inside an AI-generated answer instead of ranking a blue link. Both also point at the same underlying mechanic, an engine expanding your buyer's question into several fan-out queries and pulling sources for each one. We use "AI Search" as our umbrella term because it's what buyers actually type into Google, but we track the same thing whichever label you prefer.
ChatGPT, Perplexity, Gemini, Claude, Microsoft Copilot, and Google's AI Overviews and AI Mode, tracked through an AI visibility tool, but the mix always follows where your buyers actually search. A company whose customers live in the Microsoft ecosystem is far more likely to ask Copilot than Perplexity, so Copilot gets weighted accordingly. A consumer brand with a younger audience often sees more volume in Google's AI Overviews and ChatGPT than anywhere else. We read that from your own prompt data rather than assuming one engine matters more than another.
We find where it came from first. An answer that gets your pricing, your product names or your markets wrong is almost always repeating a source: an old page on your own site, a review profile nobody updated, a directory listing or an article from two years ago. We check the sources the engine cited for that prompt, correct your own pages first, get the third-party ones updated where we can, and keep re-running the prompt until the answer changes. Tone works the same way: if an engine keeps raising the same objection, there is usually a thread or a review behind it that is worth answering.
For the agent analytics, yes, or at least to the part of them that comes from AI bots. That can be the access logs from your server, your CDN or your hosting provider, or a small job that forwards only the lines from known AI user agents. For each request we need the time, the page, the status code and the user agent. Visitor IP addresses are not needed for this. If your CDN already keeps these logs, that is usually the quickest way in.
Faster than classic SEO in some cases, and that's the real difference. A content plan can take months to show up in Google rankings, and that's fine, it's exactly why we still run SEO: once a page ranks, it tends to keep earning for years. AI Search doesn't always follow that same timeline: when our client Heyflow's existing YouTube videos got new titles, descriptions and chapters, citations started climbing within days on content that had been sitting untouched for years. Other moves take longer to compound, our client deeploi's AI Search visibility went from 0% to about 20% over four months, and Heyflow's overall visibility climbed from single digits to over 50% across six months. It depends on how much trusted content and off-site presence already exists, and the biggest early wins usually come from work you've already done that nobody optimized for AI Search yet.
No, and anyone who promises a 100% guarantee on a ranking or visibility score is selling you something, not measuring it. What we do instead: before any engagement starts, we run a completely free strategy session using your own live data, current visibility, competitors, and where you actually stand today. Based on that status quo, our experience with similar accounts, and the specific initiatives we plan to run, we define a concrete goal for what we'll reach by a set point in the partnership, and we commit to it. We can't guarantee the outcome, nobody honestly can, but we throw everything we have at hitting it, and we stay transparent about where we stand against it the whole way through.
Alongside it, on the same team, from the same content. The work that earns rankings in Google is largely the work that earns citations in ChatGPT and Perplexity: a technically sound site, genuinely useful content, and a clear answer to the question someone's actually asking. Where the two diverge is the off-site layer, the review platforms, YouTube, and community threads AI engines lean on more heavily than classic search results ever did.
Yes, wherever the data says that's where the citations actually are. We don't run all four for every account by default. We read your prompt data first, see which sources these engines are actually pulling from in your category, and act on whichever combination, a review platform like OMR Reviews, an existing YouTube library, LinkedIn, Reddit, or outreach into the roundups your buyers already read, genuinely shows up in the answers.
