A baseline on the questions your customers actually ask, a brief for every question you lose, the pages written and published, and a re-measurement judged by a pre-registered test. Agencies charge two to five thousand a month for this with no measurement at all. We charge less, and show the before and after.
Priced on work delivered. No performance fees, no guaranteed scores — the change rule below says why.
Start with a Sprint. Move to a retainer when the first re-measurement is in. Agencies: deliver it under your own name.
Your questions, not the category's. The honest number.
The first engagement. Proof before a retainer.
One market. Specialist shops and brands.
Three markets. Brands selling across Europe.
Delivered under your name. You keep the client.
Prices in euro, excluding VAT. Pay by card (Stripe, VAT ID collected at checkout) or by invoice after intake. The first month of any retainer is the baseline month.
Most teams buy them twice — someone watching Google, someone worrying about the assistants. This is what it looks like measured together.
Seven steps, in the order the work happens. Nothing here is a slide; this is the operating procedure.
Domain, markets, three competitors, Search Console connected, a thirty-minute call for what you know first-hand.
Your own questions from Search Console, brand terms removed. Five assistants, every question twice, every market. A report with an interval on every number.
The questions you nearly win and the ones a rival holds, ranked by demand. Ten page briefs; five chosen with you.
Each page opens with the direct answer, is built like the pages that get cited — prices, figures, comparison tables, dated, bylined — and carries the facts only you have. You approve; we publish or hand over.
The cited pages that do not mention you, three to five a month, with a specific and honest ask.
Day thirty: same questions, same repeats. Per platform: changed, within noise, or new baseline. Per question: where to look next.
The next five pages. Quarterly, the question set is refreshed from Search Console and the corpora compared explicitly — never a false before-and-after.
A movement between two measurements is called a change only when a two-sided Fisher exact test finds it larger than noise, corrected across platforms so the whole report keeps a five percent false-alarm rate. Everything else is reported as within noise, in those words.
A score. An uplift. A rank. Our intervention study reports on 4 October 2026 and will say whether page changes move citations, and by how much, with an interval. Until then every proposal is priced on work delivered, and says so.
Every observation behind every number is downloadable. The method is published in three open-access studies with their data. A reproducible objection is welcome — it is how the method got better.
Ten minutes. You will see the questions we would measure you on before anything runs, and we reply within one working day with the baseline and a proposal.
Agencies: put your own domain here and name the client in the notes; we will run the client's baseline separately.