Did the campaign pay? A 30-day loyalty measurement plan in rands

The campaign generated R20000 in sales. Good news? Possibly. If most of those customers were coming anyway, the number tells you less than the dashboard suggests.

A useful loyalty report starts with a narrower question: what extra contribution did this campaign produce after paying for the campaign? Use the first 30 days to establish a repeatable method, then extend the observation window if your customers normally buy less frequently.

Agree what you are measuring

Choose one objective, such as getting recent first-time buyers to make a second purchase. Define the customers, the start date and the time allowed to return. Compare people who have had the same opportunity to buy again. A customer enrolled yesterday cannot fairly be compared with someone enrolled a month ago.

Track unique customers, qualifying visits, net sales after refunds, redeemed rewards and channel costs. Confirm that a free redemption is distinguishable from a paid purchase. A card added to a wallet is a join, not a sale.

If possible, randomly keep a comparable group out of this particular campaign. They can still receive normal service and existing programme benefits. The control helps estimate what might have happened without the extra message or offer.

Work through the rand calculation

In a hypothetical experiment, 200 eligible customers receive an offer and 200 comparable customers do not. Forty treatment customers and 26 control customers make one qualifying return purchase each. The observed difference is 14 additional purchases.

Assume each additional purchase contributes R70 after variable costs, before this campaign's incentives. That produces R980 estimated additional contribution. Suppose redeemed incentives across all contacted customers cost R300, messages cost R100 and allocated software and staff time cost R250.

Net estimated benefit is R980 minus R650, or R330. Campaign ROI is R330 divided by R650, about 51%. These are hypothetical figures, not a Futhi result. The calculation assumes comparable groups, no spillover and no other meaningful difference between them. A small test still carries substantial uncertainty.

Look beyond the first result

Check whether customers simply brought forward a purchase from next month. Review the following buying cycle before celebrating a durable improvement. Also watch complaints, opt-outs, refunds and reward misuse. A profitable short burst that damages the relationship needs attention.

Without a control, compare matched weekdays and cohorts while recording promotions, holidays, price changes and opening hours. That provides a useful directional view, but it is weaker evidence of causation.

Futhi advertises reporting on joins, repeat visits, redemptions and loyalty-linked revenue. ^[F02] Ask to see where each figure comes from and how to export what you need. You may still use a spreadsheet for the experiment.

Your first report can be one page. Give it one clear decision: continue, adjust or stop. A dashboard should help you run the business, not merely decorate the Monday meeting.