Engagement diary · Dental group · 12 locations · 14 weeks · Feb–May 2026
Twelve practices. Six were carrying the rest.
Written by Saoirse Brennan · 22 May 2026 · Figures representative, group anonymized
A 12-practice dental group spent €26,000 a month on Google Ads through one blended campaign set, at a blended ROAS of 2.6 on booked-appointment value. Fourteen weeks later the same budget returned 4.2. The diary below is the whole route, including the flat fortnight mid-migration and the one budget experiment that failed and got reverted.
2.6 → 4.2
blended ROAS, booked value
44%
of bookings arrived by phone, uncounted
31%
of budget reallocated between practices
The starting position
The group arrived convinced paid search "worked but had plateaued." The blended report supported the story: a 2.6 return on booked-appointment value, steady for two years. What the blend hid was structural. Three city-centre practices were saturated, six suburban and commuter-town practices were performing somewhere between adequately and invisibly, and three were burning budget on queries their front desks never converted, partly because nobody could see their numbers and partly because 44% of the group's bookings arrived by phone and counted as nothing. The phone finding matched the wider data, where 37% of phone leads convert during the call itself; here, receptionists were closing callers the account could not see.
Weeks 1–2
The audit table
Twelve practices, one blended campaign set, €26,000 a month. The per-branch split showed six practices booking below €60 per new patient and three above €140, with the group reading a comfortable blended average the whole time. The audit also surfaced the quiet tell: exact-match impression share above 90% at the three city-centre practices, meaning their auctions were saturated and every extra euro there bought diminishing returns.
Weeks 3–4
Measurement first
Call tracking installed per practice with booking-system imports; 44% of booked appointments arrived by phone and had never counted. Consent Mode v2 verified across the estate, since two practice sites had banners blocking Google tags outright. No media changes yet, deliberately: every later decision needed real numbers first, and the corrected baseline alone re-ranked four practices in the ledger.
Weeks 5–8
The rebuild, in tranches
One campaign set per practice: radius targeting, practice-specific landing pages with each clinic's team and reviews, shared negative lists plus local ones (rival clinic names, out-of-area suburbs, 'free dental' queries). Budgets reset from each practice's chair capacity and margin, never evenly. Carved in three tranches of four practices so learning phases staggered; the diary honestly records the two flat weeks mid-migration while bidding relearned.
Weeks 9–12
The reallocation
Controlled budget moves, 31% of total spend over four weeks, from the saturated city practices into three undervalued suburban ones whose corrected numbers beat target. Each move ran as a matched test with a written verdict before becoming permanent. One move failed honestly: a rural practice's budget increase bought impressions, not bookings, and was reverted in week 11 with the reason logged.
Weeks 13–14
The verdict
Blended ROAS closed at 4.2 against 2.6 at the start, measured on booked-appointment value both times. The spread between best and worst practice CPL narrowed from 2.9× to 1.6×, and the three rescued suburban practices now book at €54 to €71 per new patient, inside the vertical band. The monthly ledger, one line per practice, is now the group's standing board report.
What the ledger looked like, before and after
| Practice tier | Cost per new patient, week 1 | Week 14 |
|---|---|---|
| Three city-centre (saturated) | €48–€60 | €52–€63 on reduced budgets |
| Six suburban mid-table | €60–€95 (partly uncounted) | €49–€72 on corrected data |
| Three underperformers | €140+ | €54–€71 after rebuild + reallocation |
The city practices got slightly more expensive per patient and meaningfully cheaper in total: their budgets fell because full diaries cannot absorb more demand, and the freed spend is what funded the suburban gains. That trade is invisible in a blended number and obvious in a ledger.
The benchmark context
Dentists convert search at a 10.67% median on an $8.00 median CPC, with cost per lead at $72.97, per the 2026 LocaliQ benchmarks. This group's corrected numbers landed most practices inside those bands; the outliers were structural, never mysterious. The finishing ROAS of 4.2 sits above the common 4:1 health bar because dentistry carries high booked-appointment values, and we say that rather than let the number flatter us.
What generalizes
Any multi-location account reading blended numbers is subsidizing its weakest branches invisibly, and any dental or clinic group without call conversions is starving its phone-heavy practices by accounting artifact. The fix order in this diary, measurement, then structure, then reallocation, is the same one documented in the core service and applied through call and lead tracking. The free audit produces this group's week-one table for your estate, including the impression-share saturation read that decided where budget could actually grow.