1–10 Oct
2025
How it started
DentVita is a dental clinic in Prague-Vinohrady. On 1 October 2025, ahead of a first meeting, Michal Mujgos audited the clinic’s Google Ads account and emailed the co-owner 11 points in plain Czech. He did it unpaid.
Michal Mujgos:The conversion setup is essentially completely drowned out by this conversion action: Scroll page 60%, which is set as the primary conversion action. Over the last month it fired 2,516 times. […] Most of the search campaign budget, however, is being spent on non-brand and geographically irrelevant queries, which one can only assume lead to no conversions at all, or to very expensive ones. […] Personally I’d rather set the campaigns to conversions like a contact form or a click on the phone or email, not to how far down the page people scroll.
Points 2–9 and 11, summarized
- Performance Max flagged for “Health in personalized ads”
- English campaigns mixed into Czech ones
- Serious errors in the ad copy
- Stock photos, not the clinic’s own
- Subpage links mixing Czech and English
- Brand and acquisition campaigns tangled together
- Broad match triggering irrelevant clicks
- No regular negative-keyword work
- Split into more campaigns, then test
The clinic had been buying a number, not patients. The co-owner replied 9 days later.
Owners can’t audit what they’re told. Their only protection is a number that means something.
Oct – Nov
2025
6 weeks, 2 tag containers, 1 dead mailbox
To find out whether the third-party booking system’s confirmation page could be tagged, Michal made a booking through it himself.
10 Oct 2025Access promised
- Credentials by SMS
- A password in a plaintext email
- 2 competing Tag Manager containers
- Read-only access
- A cookie bar registered to a dead mailbox
- A third-party booking system
By 20 Nov 2025Measurement working
Michal’s call, in writing: “I still haven’t been given the access, but I’m not going to wait, we’ll do it on our own tracking codes.”
On 20 November 2025, unprompted, Dr. Zolarek wrote: “Thank you very much for the active approach and real commitment.”
Nov 2025
1,444 → 30.5 conversions from Oct to Nov 2025, on purpose
- ≈ −30%clicks, Oct → Nov 2025
- ≈ 2×cost per click, Oct → Nov
In November, one mixed campaign became 8 themed exact-match campaigns. The scroll stopped counting as a conversion, and the new GA4 import was only partly live.
Before the co-owner could see the drop, Michal explained the trade-off in writing: fewer clicks, but significantly better ones.
Under the old definition, October’s cost per conversion was $0.86. Under the new one, December’s was $9.19. Nothing got worse. What counted as a conversion changed.
Rendered from the account’s conversion data (Google Ads API, pulled 13 Sep 2026). Action names translated from Czech.
In plain terms: a conversion here is mostly a click on the booking button. A click, not an appointment. Completed forms and tracked calls from ads make up about 5% of conversions counted since the rebuild.
Nothing links a click to a patient in the chair, so this page reports no revenue, return on ad spend or cost per patient. Every cost per conversion here is really a cost per signal of intent.
Dec 2025 –
Sep 2026
Roughly the same conversions for 30% less money
Cost per conversion by month, USD
Dec 2025 to Sep 2026. *1–13 September only. October and November 2025 are left out: the scroll still counted, then tracking was only partly live.
The fair comparison is three full months against three, both under the new definition: roughly the same conversions for 30% less money, at a 22% lower cost per click. That is efficiency, not growth.
| Metric | Dec 2025–Feb 2026 | Jun–Aug 2026 |
|---|---|---|
| Spend | $4,795 | $3,357−30.0% |
| Clicks | 6,408 | 5,758−10.1% |
| Conversions | 520.5 | 496.9−4.5% |
| Average cost per click | $0.75 | $0.58−22.1% |
| Cost per conversion | $9.21 | $6.76−26.7% |
| Conversion rate | 8.12% | 8.63% |
The average hid a 66× spread between campaign themes.
Cost per conversion by campaign theme, 1 Oct 2025 – 13 Sep 2026
Both emergency campaigns dropped to zero impressions from July 2026, and the implant campaign was wound down between July and September.
Account CTR fell from 9.53% to 4.01% between June and August 2026. That drop is an artifact: Performance Max impressions tripled on a flat budget as Google widened placements, while Search CTR held at 15.9–17.0%.
Part of the gain comes from our own August shift toward brand and Performance Max, which captures existing demand rather than creating it.
1 Jun 2026
“We’d get nothing but acute cases”
In May, Michal proposed a page for emergency dentistry. On 1 June, the answer was not yet. Not because of cost, but because of capacity.
The constraint had moved from demand to supply, and the brief moved with it: from buying patients to recruiting a dentist. How much of that the ads caused, the account can’t say.
1–2 Jun
2026
“Were you satisfied?” Twice, 38 minutes apart
One Monday in June 2026, Michal and his girlfriend went to DentVita as patients: hygiene at 07:00, check-up at 08:00.
He had now been through the whole funnel himself, including the part after the conversion fires. Not a marketing problem. But it is the kind of thing an agency that only looks at the account never sees.
Jul – Aug
2026
Exactly 1 formal report. That gap is on us.
The work got ahead of the reporting.
| When | What happened | Written up for the clinic? |
|---|---|---|
| Mar 2026 | Formal monthly report delivered | ✓ The only one |
| Apr 2026 | Report left as an unfinished draft in Drive, every field marked PENDING | × No |
| 1 Jul 2026 | Last email the clinic received from us, about an invoice | Not a report |
| Jul 2026 | About 150 negative keywords added | × Never |
| Aug 2026 | 9 campaigns consolidated into 6 | × Never |
Both changes show up in the monthly chart above. The account has kept improving since, and nobody at the clinic has been told.
Sep 2026
Results so far
- $9.21 → $6.76Cost per conversion, Dec 2025–Feb 2026 vs Jun–Aug 2026 (−27%)
- $4,795 → $3,357Ad spend for the same periods, for roughly the same conversions (−30%)
- $0.75 → $0.58Average cost per click (−22%)
- Scroll → clickWhat counts as a conversion: a booking-button click, not a 60% page scroll
- 1 → 8Search campaigns in the November 2025 rebuild, one per theme
- Patients → hiringFrom 1 June 2026 capacity, not demand, was the limit, so the brief moved to hiring a dentist
Figures from the clinic’s Google Ads account, pulled 13 September 2026. Conversions are modeled, so counts are fractional. Amounts are converted to US dollars at the 2026 average exchange rate. Quotes from written messages, translated from Czech and Slovak, dated as shown. The previous marketer is not named.
Questions
Why did the conversion count fall from 1,444 to 30.5?
Because a 60% page scroll stopped counting as a conversion in the November 2025 rebuild, and the new tracking was only partly live that month. Under the old definition the account looked like $0.86 a conversion. Under the new one it is $6.76 (June to August 2026), and a conversion mostly means a click on the booking button.
Did the clinic get more patients?
The account can’t say: nothing links a click to a patient in the chair. Dr. Zolarek’s own impression in January 2026 was “at least 30%”, not a measured figure. By June the brief had moved to hiring a dentist.
Would this work for my practice?
If your report looks green and nobody can say in one sentence what it counts, start with the same audit. Not if your chairs are already full: don’t buy clicks you can’t seat.
In closing
A number only protects an owner if someone explains in writing what it counts and why it moved. We did the first, and the second in November, before the drop. After a single formal report in March, we stopped writing.
By June, the limit was capacity, not demand. The account can’t say how much of that was the ads.
Who this isn’t for: anyone who needs the reported number to rise every month (ours fell first, on purpose), and any practice whose chairs are already full. Don’t buy clicks you can’t seat.
In Dr. Zolarek’s words
Dr. Ivan Zolarek, DentVita:Thank you for the very good work on our website and on the marketing overall. The improvement and the response from patients is very visible. Essentially it has gone up by at least 30%, if not more.
The 30% is his impression, not a measured figure. Michal replied that the website work was “only at the beginning.”