Carrubba restaurants case study
Three fine dining rooms, one honest scoreboard.
A family of three high-end restaurants where private events carry the economics. Each property needed its own campaigns, and the family needed one monthly read that told the truth across all three.
The numbers at a glance
Proof box.
Published July 22, 2026. Updated July 22, 2026.
| Client type | Three fine dining restaurants under one family ownership, South Bay |
|---|---|
| Objective | Per-property campaigns with one truthful cross-property monthly read; grow private event inquiries |
| Channels | Google Search, Maps, Google Business Profile, paid social |
| Measurement period | Two-month inquiry trend, 2026; profile views May 2026 year over year |
| Primary outcome | Private event inquiries up 56% (10 to 14 per month across the family) |
| Secondary outcome | Profile views +136% year over year at one property, +33% at another |
| Budget context | Roughly 20x estimated return in a recent month across the family, labeled as an estimate |
| Attribution status | Inquiries tracked per property; one property's location-signal block corrected by reading site analytics alongside platform reports |
Starting condition
Three properties with distinct rooms, menus, and event economics, reported as if they were one. And at one property, a historic stone building was silently blocking location signals, so the platform's own reporting missed more than 99% of reservation activity the website could see.
The work
Per-property foot traffic and private events campaigns, paid social where it earns its keep, and a measurement fix that mattered more than any campaign: reconciling platform reports with site analytics so the family's monthly read reflects what actually happened at each address.
Deliberately excluded: blending three properties into one average that flatters the weakest room, and taking platform reports at face value where the building itself distorts them.
Methodology and limits
Inquiry counts are per property and summed only with per-property detail attached. The return figure is an estimate built from inquiry values and observed close behavior, labeled as such. The 400x measurement gap is specific to one property's location-signal problem; it is reported to show the verification method, not as a performance claim.
Results