Search split in two. One half is still ten blue links where position decides the click. The other half is a generated answer that names two or three companies and never shows a list at all. Winning one and ignoring the other leaves half your category to whoever bothered.
The uncomfortable part of AI search is that the losses are invisible. There is no line in your analytics called searches that ended in an answer about somebody else.
A buyer asks an assistant who to hire for a job in your category. It names two or three companies and cites a handful of pages. If you are not one of them, nothing in your reporting will ever tell you that this happened, and it happens all day.
Twenty years of word of mouth, a name everyone in the province recognises, a sign on the busiest road in town. A model cannot see any of it. It reads pages. Offline fame counts for nothing until it is written down somewhere an engine looks.
A security plugin, a bot filter at the CDN, or a site that only renders once JavaScript runs. Nobody in the company ever chose to keep AI crawlers out. They simply never checked, and the engines quietly moved on to a source they could fetch.
These are the sentences buyers type into Google and into an AI assistant before they ever contact anyone. Whole questions, with a decision attached to the answer.
One programme with two halves: the unglamorous technical base that moves fastest, and the published answers that compound.
Crawler access for Googlebot, Bingbot, OAI-SearchBot, GPTBot, ClaudeBot and PerplexityBot, server-rendered HTML, clean canonicals, sane redirects and pages fast enough to be fetched and indexed properly. This is where most sites are losing points without knowing it.
Organization, LocalBusiness, Service, FAQPage, Product, Review and BreadcrumbList, published consistently. It is the one place where you state the facts about your company in a form no machine can misread, and the alternative is letting an engine infer them conservatively.
The keywords buyers type into search, plus the whole sentences they type into an assistant, per market and per language. They differ more than people expect, and the second list is where categories are still wide open.
Pages built around one question each, with a short self-contained answer near the top and the evidence underneath. It reads better for humans and it survives being lifted out of the page, which is the only form a generative engine can quote.
Your company name, address, phone number, founding date and services made identical everywhere a model reads them: your site, your profiles, directories, press, partner pages. Contradictions lower confidence, and low confidence means it names somebody else.
Rankings and impressions on the classic side, and on the AI side a fixed set of buyer prompts re-run on a schedule across the engines, recording how often you are named and which competitors appear beside you. One report, both halves.
First results on the technical side within weeks. The visibility that holds takes three to six months, because it depends on published answers and consistent facts accumulating.
Where you rank today, and what happens when your category questions are asked across ChatGPT, Google AI Overviews, Perplexity, Claude and Copilot. Who gets named, which pages get cited, where you do not appear at all.
Crawler access, rendering, speed, canonicals, sitemaps, llms.txt. The half nobody enjoys and the half that moves the number fastest, usually within days of the next re-fetch.
Schema for the entities that matter: the company, the services, the locations, the reviews, the questions. Published site-wide rather than on a single page as a token gesture.
The keywords and the whole prompts, ranked by buying intent rather than volume. A question that gets asked forty times a month by people about to spend money beats one asked four thousand times by people who never will.
One question per page, answered inside the first hundred words, with the proof, the numbers and the dates underneath. Written by our own copywriters, reviewed by someone whose name goes on it.
Profiles, directories, maps, press and partner pages aligned to the same facts. Tedious, cheap, and one of the few levers that lifts every engine at once.
Dated cases with real numbers, named authors, reviews collected properly. Undated marketing copy reads as unverifiable, and unverifiable sources get left out of answers.
Rankings plus the prompt set, re-run on a schedule. You see the trend per engine next to the competitors appearing beside you, and we go after the questions where somebody else is currently the default.
They are not rivals and they are not one budget line each. They are two outcomes from largely the same foundation.
Organic and AI visibility are the channel with no media cost attached, which is why they change the shape of a business rather than just its month.
7x more leads and over €1,000,000 in extra revenue, with search and content carrying a large part of it.
“One single email campaign brought in €300,000 in extra revenue.”
Hans Josten, owner at Batisol
800 extra quote requests, from pages built to answer the questions buyers were already asking.
6,000 leads and €4.5 million sold online, with buyers arriving pre-qualified.
€490,000 in three months and the Beurs Wonen Visitor Award.
Start with the free GEO scan: it gives your site a score out of 100 across eight signal groups and tells you, in order, what to fix. Or book your free call and we will run your own category questions across ChatGPT, AI Overviews, Perplexity, Claude and Copilot while you watch.
We take on a limited number of clients per month, because the work is hands-on.
Fill in the form and we’ll get back to you within one business day.
We’ll get back to you within one business day to plan your call.