What this rate can tell you
A restaurant non-return rate measures members of a cohort who do not make an observed return within a defined window. It is often called churn, but a restaurant visit is not a cancelled subscription. No return in thirty days does not prove that someone has left permanently.
Start with the 30-day restaurant retention definition. The calculation here is its complement for an identical cohort and observation window. It is not the percentage of this month's transactions made by new customers.
Define the cohort and let it mature
A cohort can contain newly identified customers whose first visit falls in one week. Each person then receives the same follow-up period from that first visit. Someone arriving on the final day needs as much observation time as someone arriving on the first day.
Write down four choices: entry period, customer identifier, follow-up duration and qualifying return. A cancelled order or repeated scan of one receipt must not create another visit. If your source cannot distinguish two purchases from two scans, resolve that before interpreting the rate.
The formula and a checkable example
Non-return rate = (cohort customers − customers returning within the window) ÷ cohort customers × 100.
Fictional example: 50 newly identified customers form a cohort. After every member has a complete 30-day window, 15 have made at least one further qualifying visit. Retention is 15 ÷ 50 = 30%. The observed non-return rate is 35 ÷ 50 = 70%. Those 35 customers have not been labelled permanently lost.
- Returned15 customers (30%)
- No observed return35 customers (70%)
An empty cohort produces an unavailable rate, not 0%. If some purchases cannot be attached to a membership, report the identified-customer scope and its limitation. Do not silently classify unidentified visits as evidence that people did not return.
Compare like with like
Use completed cohorts, equal windows and a stable return definition. Note closures, changed opening hours and neighbourhood events. A holiday week and an ordinary week may attract different customers.
Show cohort size, returning customers and the rate together. Moving from one return out of two customers to three out of ten increases the number of returns while lowering the rate. A percentage alone hides that difference, particularly at low volumes.
| Fictional example | Cohort customers | Returns | Return rate |
|---|---|---|---|
| 1 out of 2 | 2 | 1 | 50% |
| 3 out of 10 | 10 | 3 | 30% |
Turn the observation into a focused test
Change one thing at a time: explain the reward more clearly, improve the programme's presentation or address a service issue. Then observe another complete cohort. An improvement is still an association; seasonality, a different customer mix or an unrelated promotion might explain it.
For the programme itself, read the digital restaurant loyalty guide. Minerva Flow tracks identified visits and programme results. A new membership or awarded point is not, on its own, evidence of attributable additional revenue.
