Restaurant reporting often compresses several customer steps into one number. That makes a campaign look simpler, but it makes decisions harder. The right approach is to keep each action useful by preserving its definition.
Start with the customer path
| Step | Evidence available | Do not rename it as |
|---|---|---|
| Search or listing view | The restaurant appeared in a result | Visit or customer |
| Direction request | A route was requested | Arrival or order |
| Call action | A call was initiated | Connected call or reservation |
| Order or reservation click | A destination was opened | Completed order or booking |
| POS / booking record | A transaction or reservation exists | Incremental result caused by one channel |
A dated example: Burrito Splendido
From 1 July to 14 August 2026, Google Ads reported 1,384 actions for C$2,915.70, or C$2.11 per action after rounding. The actions were predominantly direction requests. The prior same-definition period, 17 May to 30 June, recorded 1,198 actions and C$3,105.03 spend.
The period comparison is +15.5% actions with 6.1% lower spend. It is not a controlled causal experiment and does not prove restaurant visits, orders or revenue. Data from 15–17 August is excluded because local actions can report with a delay.
Read the full case and definitions →
A restaurant measurement checklist
- Write down the exact action name and the platform that reported it.
- Keep the reporting dates, currency and attribution setting beside the result.
- Separate location-level direction, call and site actions.
- Connect reservation, order and POS data only when the identifiers and timing support the link.
- Report unmatched or ambiguous records instead of forcing attribution.
- Use accepted enquiries, completed reservations and paid orders as separate downstream outcomes.
A useful direction request can still be an important optimization signal. The discipline is to value it as a direction request.
Use the result to choose the next question
If actions rise but orders do not, inspect the location page, menu, ordering path, opening hours and the offline experience. If actions are low, inspect visibility, relevance and paid-search intent. The metric tells you which question to ask; it does not answer every question.