A random share of customers
A small share of customers does not get the promotion and the rest do. The real effect becomes visible.
A store owner asks, "Does the program actually make money?" To work out loyalty program payback you do not need industry percentages. You need four of your own numbers and an honest comparison with people who are not in the program. Below, loyalty program ROI is broken into steps; the result depends only on your data, and we promise nothing in advance.
What costs and benefits consist of
Costs
- discounts and bonuses actually given;
- the system, the till, the app, the messages;
- staff time and marketing;
- card reissue, if there are cards.
Benefits
- more visits from the same customers;
- a higher average receipt;
- customers who come back instead of leaving;
- data: you can see whom to offer what.
The main trap is to count all revenue from cardholders as a benefit. Those customers would have bought anyway, so only the increment belongs to the program.
The formula
1. The baseline
Take a group of customers who receive no offers (the control group) or the period before launch. Calculate the average number of visits, the average receipt and the margin.
2. The program result
The same indicators for program participants over the same period.
3. Margin increment
Participants' margin minus the margin of a control group of the same size.
4. Subtract the cost of discounts
Discounts and bonuses reduce margin, not revenue: subtract them from the increment.
5. Subtract the program costs
System, messages, team time.
6. Compare
If the remainder is positive, the program pays back within the chosen period. If not, change the mechanic, not just the advertising.
In short: effect = (participants' margin − discounts) − control group margin − program costs.
Worked example (made-up numbers)
Example conditions: 1000 customers in each group, an average receipt of 150 lei, a 25% margin, one month. The control group makes 2.0 visits on average, program participants 2.4. Discounts and bonuses cost 3% of participants' revenue, and program costs are 3000 lei per month.
| Indicator (example) | Control group | Program participants |
|---|---|---|
| Visits per customer per month | 2.0 | 2.4 |
| Revenue | 300,000 lei | 360,000 lei |
| Margin at 25% | 75,000 lei | 90,000 lei |
| Discounts and bonuses (3%) | 0 | 10,800 lei |
| Margin after discounts | 75,000 lei | 79,200 lei |
Increment: 79,200 − 75,000 = 4,200 lei. Subtract program costs of 3,000 lei and 1,200 lei per month remain. It pays back, but only just.
Now suppose frequency grew not to 2.4 but to 2.2. Revenue 330,000, margin 82,500, discounts 9,900, remainder 72,600. That is 2,400 lei less than the control group, before program costs. The program loses money.
The example shows sensitivity, not a forecast. Under these conditions the break-even point is about 2.36 visits per customer. Put in your own numbers: your threshold will be different.
One more remark about margin. If you do not know the margin by product, take the store average: it is enough for a first estimate. Precision matters later, when you start comparing different mechanics with each other. And remember that a discount on a low-margin product eats profit faster than one on a high-margin product, which is why the same percentage across the whole shelf is rarely optimal.
How to get a control group
Without a control you cannot tell the program's effect from the season, roadworks on the next street or a price rise. Options:
Twin stores
Similar outlets where the promotion runs in one and not in the other.
Before and after
The weakest method: any event of the period lands in the result. It works if there is no other option.
Before launch decide which indicator means success and over what time. If you define it afterwards, the result becomes a fit to the data.
Remember the flip side too: a program can pay back overall and still lose money in some segments. Split the result by customer groups, for example new, regular and lapsed. It often turns out that a discount for regulars adds almost nothing, while an offer to those who have started to drift away works better. Then the money is worth moving to where the increment is clearer.
Measurement checklist
- A control group or a baseline period is defined
- Visits, average receipt and margin are recorded before launch
- Discounts and bonuses are counted as actually given, not as planned
- Bonuses with a future expiry date are counted as a liability
- System and messaging costs are included
- The result is reviewed at least once a quarter
What to watch besides margin growth
Margin growth is the main indicator, but it arrives with a delay. To notice a problem earlier, watch intermediate signals:
Visit frequency
Whether it grows among participants and in which segments. If it grows only among people who came often anyway, the program hands out discounts for nothing.
Share of customers who came back
How many shoppers who made a first purchase on the card returned a second time. It is the most honest sign of a living program.
Share of receipts with a discount
If almost every receipt has a discount, it is no longer an incentive but a price cut for everyone.
If the signals are poor, do not wait for the end of the quarter: change the mechanic on a small group and compare again. More about typical slips in the article on loyalty program mistakes.
What Synngate Loyalty offers
The ERP module has a built-in promotion control group: part of the customers do not get the offer and the real effect is visible. The web center has a promotion effect report that compares the promotion group with the control group. Lifecycle analytics shows visit frequency and the segments from which the baseline is built, and mechanics let you change the discount size without a programmer. You can compare the ERP module and the web center or request a demo on your own data.
Frequently asked questions
What payback should a program have?
There is no single figure. It depends on margin, purchase frequency and discount size. Calculate with your own data, as in the example above.
What if there is no control group?
Compare the period before and after launch and set it against a similar store with no program. The result will be less reliable.
Should unspent bonuses be counted?
Yes, as a liability. If bonuses expire, part will disappear, but it is more prudent to plan as if customers will spend them.
When should we look at the result?
Long enough for customers to come back a few times. For a store with frequent purchases that means months, not weeks.