Loyalty Program Analytics: Data, Measurement and Reporting

Loyalty program analytics is the practice of collecting member, transaction and engagement data, analyzing it to understand how members behave and what the program changes, and using the results to decide what to keep, fix or stop. It covers the data, the measurement method, the reporting rhythm and the decisions that follow.

Enrollment counts flatter most programs. Deloitte's 2025 survey of 5,564 US loyalty members found that "the average consumer enrolls in eight loyalty programs, yet actively participates in only five" (Deloitte, 2026). Counting enrolled members tells a program little; analytics is about finding out which members are active, why, and what the program does for them. This guide covers how to set that up. Metric definitions and formulas are in the loyalty KPI dashboard, and common obstacles are in the guide to loyalty data analytics challenges.

Key Takeaways

What questions should loyalty analytics answer?

Analytics should answer the questions behind program decisions: who is active, what the program adds, which offers work, who is likely to lapse, and what the program costs.

Write the questions down before building dashboards, and fix definitions such as "active member" (for example, a purchase in the past 90 days) before the first report, because changing a definition later changes every trend. A dashboard without a decision attached tends to be read once and ignored.

What data does loyalty analytics need?

Transactions, member profiles, engagement, redemption and campaign data, joined into one member record that every report draws from.

• Transactions: what each member bought, where, when and at what margin, plus non-member transactions for comparison.
• Member profile: enrollment date, tier, preferences and consented contact details.
• Engagement: logins, app activity, email and text responses, non-purchase actions such as reviews or referrals.
• Redemption: what members redeemed, when, and points earned but not yet redeemed.
• Campaigns and offers: which member received which offer, and whether they acted.
• Service: complaints, returns and support contacts.
• Partner data: for coalition or channel programs, the transactions partners report.

Match each source to the member through a common identifier, such as a member ID linked to email and payment or receipt data, and check the match rate: transactions that can't be tied to a member are invisible to the analysis. Clean duplicates and keep formats consistent across sources, so a member who appears twice is not counted as two. The challenges guide covers data quality and integration problems in more depth.

How should you segment and analyze members?

By segment and by cohort: group members by value, behavior and enrollment date, and compare how each group changes over time instead of reading program-wide averages.

Program-wide averages hide the story. A rise in average spend can come from a few large members while most of the base drifts away. Segment members by value (spend and margin), behavior (frequency, recency, categories bought, channel) and stage (new, active, lapsing, lapsed). Then track cohorts: members who joined in the same month, followed over their first year, show whether onboarding and early offers keep people active.

Brandmovers' program for Signia, an audiology manufacturer, shows segmentation in a B2B setting. The program on BLOYL™, Brandmovers' enterprise loyalty platform, uses a dynamic segmentation model that classifies Signia's customers into groups including buying groups, small and midsize businesses, family offices and independent providers, with promotions and rewards tailored by tier, purchase behavior and engagement level. Members of the program, called Aspire, recorded 15% unit growth in 12 months and an 87.3% average engagement rate on a recurring basis (disclosed by Brandmovers; Signia case study). The case reports no comparison group, so the figures show the program in use rather than a measured lift.

How do you know whether the program is working?

Measure what the program adds, not what members do: compare members with a holdout, a staggered rollout or a matched set of non-members.

Members usually outspend non-members, but that gap is not proof of a program effect, because the customers who enroll were often already the most engaged. One Brandmovers example shows the problem. A Canadian regional distributor's points program for its smaller customer accounts, on BENGAGED™, recorded a 25% average sales increase among enrolled customers versus 5% among non-enrolled customers (disclosed by Brandmovers; distributor case study). Customers were not randomly assigned to enroll, and the case does not state the period measured, so the gap may reflect a program effect, differences in who enrolled, or both; it is not a measured lift.

Three methods give a cleaner answer:

1. Holdout group. Randomly keep a share of eligible customers or members out of the program, an offer or a campaign, and compare them with those who received it over the same period. For an established program, holding members out of individual offers or changes is usually more practical than holding customers out of the program itself.
2. Staggered rollout. Launch the program or a change in some regions or stores first, choosing them before looking at results, and compare them with places that start later. Watch for customers who shop in both groups, which blurs the comparison.
3. Matched comparison. Where a holdout is not possible, compare members with non-members who looked similar before enrollment, such as in spend, frequency and tenure. This is weaker, because unmeasured differences remain.

Before a test starts, set the measure that will decide it, such as second-purchase rate or incremental margin, and the minimum group size needed to detect a difference large enough to matter. Run tests long enough to cover a full purchase cycle, and treat small differences between small groups as likely noise. In B2B programs with a few hundred accounts, a handful of large accounts can swing the result, so compare accounts of similar size or report results with and without the largest accounts. The guide to loyalty program ROI shows how to turn the incremental result into a return figure.

How do NPS and customer effort scores fit into loyalty analytics?

Use Net Promoter Score and Customer Effort Score to help explain member behavior, not to replace it: link survey answers to each member's purchases and engagement.

Behavioral data shows what members do; short surveys can help explain why. Net Promoter Score asks "How likely are you to recommend us to a friend or colleague?" on a zero-to-ten scale, and the score is "the percentage of customers who are promoters (those who scored 9 or 10) minus the percentage who are detractors (those who scored 0 to 6)" (Bain & Company). Customer Effort Score asks how easy it was to get something done, such as resolving a service issue. In a study of "more than 75,000 people interacting with contact-center representatives or using self-service channels," the authors of a 2010 Harvard Business Review article reported that the Customer Effort Score is "a better predictor of loyalty than customer satisfaction measures or the Net Promoter Score" (Harvard Business Review, 2010). That research looked at service interactions, so effort scores are most useful at points such as enrollment, redemption and support.

Three practices make survey scores useful in a loyalty program:

• Link responses to the member record, so scores can be compared with later purchases, redemption and lapse rather than read as a standalone number.
• Look for mismatches. A member who says they would recommend the brand but has stopped buying, or a frequent buyer who reports high effort, is worth investigating.
• Allow for response bias. Members who answer surveys may differ from those who do not, so compare respondents' behavior with non-respondents' before generalizing.

How can predictive analytics help?

Predictive models score which members are likely to lapse, buy again or respond to an offer, so the program can act before a member goes quiet.

• Churn risk: members whose purchase or engagement pattern resembles members who previously lapsed, so the program can intervene early.
• Propensity: members likely to buy a category, respond to an offer or move up a tier. A high score is not the same as being influenced by an offer, because some of these members would buy anyway, so test offers against a holdout before spending on them.
• Lifetime value: the expected value of a member over time, so acquisition and retention spending can be weighed against it.

A new program may lack enough lapse history to train a model; until it has a year or so of data, a simple rule such as no purchase in twice the usual interval can flag at-risk members. Treat scores as a way to prioritize, not as certainty. Test any action a model triggers against a holdout, because a member flagged as at risk may have returned anyway, and refresh models as behavior changes.

Illustrative example: a churn rule flags members whose time since last purchase is twice their usual interval. The program sends half of the flagged members a reminder offer and holds out the other half. If the offer group returns at no higher rate than the holdout, the offer costs money without changing behavior and is cut.

How often should you report on a loyalty program?

On a set rhythm matched to the decision: daily or weekly for operations, after each campaign, monthly for program health and quarterly for economics.

Finance needs the points outstanding and the expected redemption behind the liability; the guide to loyalty program liability covers what CFOs look for.

How do you turn findings into program changes?

Treat each finding as a hypothesis: design a change, test it against a control, decide on the result, and roll out only what works.

Illustrative example: monthly reporting shows that members who joined in the past quarter redeem less than earlier cohorts, and a survey suggests they find the first reward too far away. The team tests a lower first-reward threshold with half of new members and keeps the other half as a control. After a full purchase cycle, the test group's second-purchase rate and total spend are compared with the control's, so a purchase pulled forward is not counted as a new one, and the change is rolled out only if the gain covers the extra reward cost. The same loop applies to reward mix: if redemption data suggests members prefer experiential rewards to discounts, check whether point pricing or catalog placement explains the pattern, then test adding more before reshaping the catalog.

Record each test, its result and the decision. Over time, the log shows what works for your members, which is worth more than any single dashboard.

What privacy rules apply to loyalty analytics?

State privacy laws govern how loyalty data is used; in California, businesses offering benefits in exchange for personal information must give a notice of financial incentive.

California's CCPA regulations require businesses offering a financial incentive or price or service difference to give a Notice of Financial Incentive that explains "the material terms" so "the consumer may make an informed decision about whether to participate", including "the categories of personal information that are implicated" and "the value of the consumer's data" (Cal. Code Regs. tit. 11, section 7016). The notice must be available before members opt in, so plan new analytics uses of member data before they are disclosed, not after. Several other states have their own comprehensive privacy laws, so check the rules in each state where members live. Collect only the data your analysis needs, tell members how you use it, and honor opt-outs. This is general information, not legal advice.

What tools and team does loyalty analytics need?

A loyalty platform with reporting, one member record, a BI tool, and people who own the questions and act on the answers.

BLOYL supports analytics with real-time dashboards, A/B testing against a control group, predictive churn analytics, financial performance tracking, exports to Tableau and bidirectional CRM and customer data platform data flows. Whatever the tools, assign an owner for each recurring report and each test, and train the marketing team to read cohort and test results, not only totals. Analytics only grows a program when someone is responsible for acting on it.

Conclusion

Loyalty program analytics helps grow a program when it answers specific decisions and leads to tested changes. Define the questions, join the data into one member record, analyze by segment and cohort, measure what the program adds with a holdout or staggered rollout rather than a member versus non-member gap, use predictive scores to act early, report on a set rhythm and turn every finding into a tested change.

Sources

• Deloitte Insights, "Reshaping loyalty programs in an era of value seeking" (2025 Consumer Loyalty Program Survey)
• Bain & Company, Net Promoter System, "Measuring Your Net Promoter Score"
• Matthew Dixon, Karen Freeman and Nick Toman, "Stop Trying to Delight Your Customers," Harvard Business Review (July to August 2010)
• Legal Information Institute, Cal. Code Regs. tit. 11, section 7016, Notice of Financial Incentive