The KPIs of a loyalty program must show whether customers enroll, use benefits and return to purchase, as well as help assess the cost of that operation. Sign-ups and rewards issued are part of the monitoring, but they are not enough to demonstrate return.
To turn indicators into decisions, define who is included in each calculation, which period will be observed and what counts as a purchase, activation or redemption. Below you will find operational formulas and hypothetical examples. The numbers serve to explain the calculations; they are not market targets or Smartbis customer results.
Define the base before calculating the indicators
A rate is only comparable when numerator and denominator describe the same population. Before setting up monitoring, record four criteria:
- Identification: how to recognize the same customer across different purchases and handle duplicate registrations.
- Valid transaction: which orders enter the calculation and how cancellations and returns will be treated.
- Window: purchase period or number of days after enrollment that will be observed.
- Group: which customers or vouchers belong to the analysis, without changing the composition in the middle of the comparison.
If buyer identification is incomplete, report the coverage. A rate calculated only with identified customers does not necessarily represent all company sales. When the denominator is zero, present the indicator as not calculable, instead of assigning a 0% rate.
Enrollment and activation: does the customer join and start participating?
Enrollment rate
Enrollment = new sign-ups among eligible invited customers ÷ unique eligible invited customers × 100.
This definition measures invitation conversion. If 200 eligible, not-yet-enrolled customers received the invitation and 50 enrolled within the defined period, the rate is 25%. Count each person once, even if they receive more than one message.
If you do not record invitations, you can measure participation among sign-ups within the identified purchaser base. Give this metric a different name and report the denominator: it answers a different question.
Activation rate
Activation = enrolled members of the group who performed the defined action within the period ÷ total enrolled members of the same group × 100.
Choose an action meaningful to the program, such as the first eligible purchase after enrollment. In a group of 50 new enrollees, if 30 perform that action within 30 days, activation will be 60%. Wait for the full 30 days for all participants before comparing the result with another group.
Automatic credit at registration and a completed purchase are different events. If credit is used as activation, make that explicit so a system action is not interpreted as a customer-initiated action.
Redemption rate: which benefit was actually used?
“Redemption rate” can refer to vouchers, people, points or monetary value. Choose the unit before calculating. The table presents three distinct readings:
| Indicator | Formula | Main caution |
|---|---|---|
| Use of vouchers from a batch | Vouchers used from the batch ÷ vouchers issued in the batch × 100 | Monitor the same set until the cut-off date. |
| Participants with a redemption | Customers in the group with at least one redemption ÷ eligible customers in the group × 100 | Count people, without adding their redemptions as new customers. |
| Use of the value issued in vouchers | Value actually used from the batch ÷ value issued in the same batch × 100 | Use the same currency and document partial uses and adjustments. |
Assume a batch has 100 vouchers issued and 35 used by the cut-off date: the usage rate will be 35%. If the 35 uses come from 28 people, that does not mean 35 customers redeemed. And if vouchers have different values, the quantity-based rate will not necessarily equal the value-based rate.
Record cancellations and reissuances separately and maintain a consistent rule for including or excluding them from the denominator. A reissue of the same benefit should not silently inflate the total.
Avoid dividing all redemptions in a month only by vouchers issued in that month: part of the uses may come from earlier issuances. That quotient measures flow and can even exceed 100%; it is not the conversion rate for that batch.
Balance still available does not equate to a benefit definitively lost. The term breakage usually designates the portion of benefits that will not be used, but a pending balance does not allow that conclusion. In Smartbis, cashback balance does not expire; a voucher may have a validity period according to configuration. The cashback page clarifies this difference.
Repurchase, frequency and retention: who came back?
Second-purchase rate
Second purchase within 60 days = customers from a first-purchase group who bought again within 60 days ÷ total customers in that group × 100.
If 40 of 200 new buyers returned within that interval, the rate is 20%. The 60 days are an illustrative window: adapt it to the business cycle and wait for the full observation. Do not compare customers tracked for two months with others who arrived a week ago.
Frequency and average ticket
Frequency = valid purchases in the period ÷ unique customers who purchased in the period. With 300 purchases from 120 buyers, frequency is 2.5 purchases per buyer.
Average ticket = revenue from valid purchases ÷ number of those purchases. If the same 300 purchases generated R$ 24,000, the ticket is R$ 80. Define how discounts, returns and shipping are treated and keep it consistent across periods.
Because the frequency above considers only those who purchased, it can increase even when part of the base stops returning. Analyze it alongside retention. The article on purchase frequency in loyalty explores the role of the interval between visits in more depth.
Retention by customer group
Retention = buyers in the initial period who also bought in the subsequent period ÷ buyers in the initial period × 100.
If 45 of 120 buyers in one month returned the following month, that group's monthly retention was 37.5%. New customers in the second month are excluded from the calculation. In businesses with long intervals between purchases, a monthly window may be uninformative.
Absence of purchase in the window indicates inactivity according to the chosen criterion; it does not prove definitive customer loss. Compare groups with similar tenure, channel and purchase profile.
CAC, LTV and ROI: is the program economically sustainable?
Customer acquisition cost
CAC = acquisition expenses attributed to the segment ÷ new customers acquired in that segment. For example, R$ 4,000 to acquire 100 buyers represents a CAC of R$ 40, provided expenses and customers have compatible attribution.
Enrolling an existing buyer in the program does not make them a newly acquired customer. Separate acquisition, enrollment and reactivation so as not to artificially reduce CAC.
Customer lifetime value
LTV can be calculated on different bases, such as revenue or margin. State which one you are using. For an initial analysis, track the cumulative contribution margin of a group ÷ number of customers who entered that group, including those who stopped buying during the period.
This calculation shows value observed up to the cut-off date. It does not, by itself, represent the entire future value of the relationship. Projecting LTV requires assumptions about retention, frequency, costs and margin; if there is insufficient history, present scenarios rather than a definitive number.
Return on investment
One possible managerial reading is: ROI = (incremental contribution margin before program costs − program costs) ÷ program costs × 100.
“Incremental” means the additional portion you estimate was caused by the program. If that additional margin is R$ 3,000 and the corresponding costs are R$ 2,000, the estimated ROI will be 50%: (3,000 − 2,000) ÷ 2,000 × 100.
Include granted benefits, platform, communication and operations in the appropriate place in the account. If the cost of a reward has already been deducted from margin, do not subtract it again. To organize cost components, the guide on investment in loyalty programs offers a complementary view.
Sale with benefit is not proof of incremental sale
A purchase that used a voucher is associated with the program. This does not prove that the person would not have purchased without the incentive. Comparing enrollees with non-enrollees can also distort the reading: customers who already purchased more may be exactly those most interested in enrolling.
When feasible, evaluate a new campaign with comparable groups defined in advance and a common observation window. Randomly assigning the campaign among eligible customers can help reduce profile differences while preserving promised benefits. Compare the result per eligible customer, not only among redeemers.
If only a before-and-after comparison is available, record factors such as seasonality, price changes, product availability and other campaigns. The result will be indicative, with lower confidence for attributing the effect to loyalty.
How to turn KPIs into a decision routine
| Observed combination | Next investigation |
|---|---|
| Enrollment increases while activation remains low. | Check the promise in the invitation and the difficulty of the first action. |
| Issuance increases and voucher use declines. | Review timing, rules, communication and possible fulfillment failures. |
| Redemptions increase and margin falls. | Examine reward costs, combined discounts and purchases that would have occurred anyway. |
| Ticket rises and retention falls. | Check whether a few high-value buyers are hiding the exit of other customers. |
Use frequent operational reviews to identify failures and full periods to evaluate repurchase and retention. Define goals from history and the business's economic capacity. A rate from another segment, with different rules and windows, is not a sufficient reference.
Smartbis provides dashboard and return reports to monitor customers, benefits, vouchers, redemptions and recurrence. The formulas in this guide are analysis methods; they do not mean that all are calculated automatically in a native panel. Confirm the available fields and complement the evaluation with the company's sales and cost data.
Sources consulted
Consulted on October 10, 2026. Product information was checked on Smartbis's official pages about return reports and cashback and voucher rules. The operational formulas were made explicit with their denominators; the numerical examples are hypothetical and do not constitute benchmarks.
To assess how to monitor benefits and repurchase in your company, learn how Smartbis works.