Gramedica: a 90-day report

An experienced physiotherapist was being found for his own name, and for almost nothing else. The treatments he is good at were not what his website was being found for.

The site went live on 6 May 2026. This is a report on what the instruments recorded through 30 July 2026 — the first 85 days.

The diagnosis

The problem was visible from outside before anything was built. Search Console for February–April 2026 shows it plainly: the old site earned its impressions on the practice's own name and very little else, across 19 indexed URLs (snapshot 6 May 2026). Someone searching for a physiotherapist in Poznań, rather than for this physiotherapist, did not find him.

That diagnosis is the transferable part of this engagement, and it is worth separating from the build. It came first, and everything after it was execution.

Gramedica is a client build and the terms are not disclosed; the case study covers the reasoning, and this report covers the numbers.

What we did

We rebuilt the site. The search work that went with it was baseline, not a campaign: correct indexing, a page per service, clean titles, structured data. No link building, no content programme, no ongoing optimisation.

Focused search work has not been attempted. This is what baseline work recorded in 85 days, for one practice.

The rebuild was not the only thing that changed. In the same period the practice was repositioned and a premium service launched: we wrote the strategy, specified the offer, and built the page it sits on; Paweł approved it and delivers the service itself. So the numbers below are not attributable to a new website alone. What they measure is a bundle — a new site, a sharper position, and a new service, shipped together.

What moved

One query moved all the way through to clicks.

GA4 independently shows the klawiterapia service page as the most-read page after the homepage. Two instruments that do not talk to each other agree, and the rebuild is what gave that treatment a page of its own.

That is a niche term. It was the winnable one: specific, with real intent behind it, and reachable inside 85 days. Interpretation: prioritising this service may have contributed to the improvement. The figures do not isolate its effect. The broad terms are a different fight, and further down this report says where they stand.

Before, the practice could only count who looked at the contact page. Now the site records clicks on the WhatsApp and phone links, and Search Console shows which searches bring people to it. That is a capability claim, not a growth claim.

What the new instruments recorded.

One of those counts clicks, the other counts distinct users, and we will not blur them.

Behind those arrivals, the search picture widened. Comparing February–April against May–July, non-branded impressions rose 33% (3,142 → 4,175) through the low season, and more than 15 queries appeared from zero at 30 impressions or more.

Where people did arrive from search, they stayed.

What it means

In this reporting window, some queries gained impressions without clicks, while the narrower service query recorded more clicks.

Both ends are visible in this one report, and the contrast is the lesson. klawiterapia poznań sits at position 3–4 and the clicks are arriving. fizjoterapia poznań sits at about 40, impressions growing, no clicks at all. One was a winnable term and it has been won; the other sits at about 40 with no clicks, and is a different contest. The diagnosis helps us choose which search opportunities to pursue first.

What is still open

Three things have not moved yet. They are the next phase of the work.

Total visits are flat. Not up. The composition changed — different people, different queries — but the volume did not, and 85 days is too short to claim otherwise.

Contact-page visits roughly halved. The new site does not route people through a contact page; the messaging button is on every page. The step we used to measure got shorter, so the proxy fell, and the new contact actions have no old-site counterpart to compare against. At most 19 people reached a contact page at all — an upper bound across three contact URLs that may double-count anyone who visited more than one (GA4, 6 May – 30 July 2026).

The broadest money query is still at position ~40, with zero clicks. The caveat above states it in full. Impressions are the leading indicator; they are not the result.

Paweł's own account adds what the instruments cannot see: he reports that people called, that some became patients, and that one Pro Session package sold.

The next measurement is October 2026, on the same instruments and the same windows.

How we measure

The labels below are part of the result. They separate a count from an upper bound, a measured event from a missing baseline, and visibility from an outcome.

The instruments, and what each one is allowed to say.

• Contact actions — GA4, 6 May – 30 July 2026. WhatsApp clicks and distinct users are reported separately, never merged.
• Organic quality — GA4, Poland-filtered. Engagement rate and average time describe the quality of a visit, not the volume of traffic.
• Search visibility — Google Search Console. The old-site snapshot recorded 19 indexed URLs on 6 May 2026. Impressions count how often a page was shown, not how often anyone clicked.

What this report does not claim.

• The old site had no event tracking, so this report does not calculate a before-and-after change in contact actions. It reports the change in what could be counted at all.
• 48 clicks are not 48 people. The measured count is 31 distinct users.
• More impressions are not more visits. Total visits were flat to slightly down over the same window, while contact-page visits roughly halved.
• The results belong to a bundle — rebuild, repositioning and a new service, shipped together — and no instrument here can apportion them.