Measuring loyalty programme impact

Judge a loyalty programme by the behaviour and contribution it changes, not by sales booked to member accounts. First set the outcome, eligible customers and review period. Then assess the result against a credible comparison and the cost of delivering the benefits.

Define the outcome and records

Pick a primary outcome that fits the buying cycle — a second purchase or retained purchasing among a defined group, for example. Record when customers enter the group, which orders count, how refunds are treated and when observation ends. Keep those definitions stable throughout the review.

Track enrolment, eligible purchases, reward claims, completed uses and contribution separately. Enrolment shows take-up; reward use shows whether members receive a benefit. Neither establishes that the programme caused extra purchasing.

Link customer eligibility and enrolment dates to orders, refunds and reward events. A claimed voucher may be used later or never used. Reconcile totals with source transactions, and state whether reported revenue includes discounts, delivery charges and GST.

Late refunds can change a period’s result. Missing product costs or unmatched store purchases should remain visible as data gaps.

Operationalise the indicators

Give each indicator a written calculation, including its numerator, denominator, customer population and period. For customer retention rate, record customers at the start and end of the period and new customers acquired during it. The formula is (ending customers − new customers) ÷ initial customers × 100.

Repeat purchase rate describes the percentage of the customer base who are repeat buyers; retention rate instead uses the starting customer count and accounts for newly acquired customers. Keep the chosen calculation and eligible population consistent between reviews. A movement in one measure does not automatically mean the other changed in the same way.

Key Loyalty Programme Impact Metrics

Compare outcomes fairly

Use a credible comparison where feasible, choosing an approach suited to the rollout and available data. For detailed comparison procedures, including assignment and observational methods, see the supporting article. Report the comparison’s assumptions and limitations when reporting estimated programme effects.

Preserve extract definitions

Keep a record of each data extract used in a review: when it was created, the date range covered, the report type and whether it processed successfully. If an export is still running, record that it is in progress rather than treating its incomplete results as final. A failed export should prompt a new extract, not be interpreted as zero activity.

Export column headings and formats can vary: fields may use standard labels, snake case or legacy names, and headings can be renamed. Retain the field mapping and format used for each review so that a later comparison is not affected by a changed label or a different selection of columns. Include only fields needed for the stated measures, while keeping the definitions clear.

Calculate the value of the change

On a consistent GST basis, compare contribution per eligible customer across groups. Define contribution as receipts after discounts and returns, less the costs that change with those orders. Scale the difference to the relevant tested population, then deduct additional programme costs absent from those order figures, such as software, service or administration.

Map each reward cost to the calculation. If net receipts already include a reward discount, subtracting its face value again counts it twice. A free item may instead add a supply or fulfilment cost outside the order record. State which costs were measured, estimated or missing.

Read the financial estimate beside benefit delivery: who qualified, who tried to use a benefit, who completed a use and what remains unresolved. Report uncertainty and data gaps with the result. The decision to continue or revise should reflect the intended behaviour, member experience and estimated additional contribution for the customers and period actually assessed.

Interpret reported sales consistently

A sales figure can include transactions that do not match the outcome definition. In Shopify sales reports, gross sales are product price multiplied by quantity before taxes, shipping, discounts and sales reversals; pending, cancelled and unpaid orders are included, while test and deleted orders are excluded. Check report definitions and filters before using a figure as evidence of completed purchasing.

Keep the reported measure distinct from the measure used in the impact calculation. If the review concerns eligible customers’ completed purchases, explain how report totals were filtered or reconciled to that population and outcome. Do not present a change in a platform’s gross-sales figure as a change in the programme’s measured contribution unless the underlying definitions match.

Sales Report vs. Impact Calculation Definitions