Define the decision first
An evaluation should support a decision: keep the rule, change it or stop. For your 2027 programme, write down what evidence would justify each option before looking at results.
Thirty days can uncover journey problems and provide early observations. They do not guarantee complete retention data or proof of incremental sales.
Days 1–3: agree on definitions
Define an eligible paid visit, an identifiable guest, the return window and cancellation treatment. Decide who may change programme rules and who may correct points.
Prepare a record of cohort start dates, eligible guests, returning guests, rewards used, direct costs and incidents. Keep definitions stable. If the rule changes, record when and separate the periods.
Days 4–7: rehearse with the team
Complete enrolment, a purchase, points earning, redemption and a cancellation. Use a guest’s phone and the actual staff device. Check permissions, displayed balances and history.
Also test mistakes: the wrong guest, a repeated action or an ineligible purchase. Write down the correction process. The Minerva Flow fictional demo explains the journey but does not replace a test with your real configuration.
Days 8–21: observe normal service
Present the same rule to eligible guests under the published conditions. Record recurring questions without gathering unnecessary personal information.
Review sign-ups, subsequent visits and redemptions. Keep cancelled events separate. Document any other promotion introduced during the trial because it will affect how you interpret changes.
Days 22–30: separate signals from conclusions
| Observation | What it shows | What it does not establish |
|---|---|---|
| New cards | Enrolment adoption | Additional visits |
| Returning guests | Returns within the defined window | Returns caused by the programme |
| Redeemed rewards | Actual use of benefits | Full profitability |
| Member sales | Purchases observed in that group | Incremental sales attributable to loyalty |
Use a consistent 30-day return-rate definition. A guest enrolled on day 29 has not had thirty days to return by day 30. Separate incomplete cohorts and plan a later review.
Connect reward costs to contribution
Fictional example. An assumed additional purchase leaves $8 after the variable costs included in your model. If fixed programme costs are $80 for the period, ten genuinely additional purchases would cover them under those simplified assumptions. If reward costs were not already deducted from the $8, they must be added separately.
The key word is additional. Purchases that would have happened anyway are not a new gain. The calculator below models an assumption; it does not attribute revenue or report an observed result.
Decide what happens next
Fix journey errors first. If the rule is understandable, costs remain acceptable and the observations are usable, define the next review period. Otherwise simplify the reward or adjust the threshold while clearly announcing changed conditions.
A suitable comparison group can strengthen the evaluation, but it needs planning before a causal claim is made. Holidays, weather, events and which guests choose to join can all change the numbers. Do not turn a handful of returns into a growth promise.
For tool selection, see the 2027 loyalty comparison. For coffee shops, adapt the neighbourhood café reward guide.
Method. Proposed by Minerva Flow on October 28, 2026. Timelines, thresholds and calculations are planning examples. No study or customer result is reported; use your actual costs and an appropriate evaluation method.