Before launching a loyalty program, name the behavior it should change
A loyalty program can spend money on purchases its best customers would have made anyway. Its investment case therefore depends on incremental value: purchases or other benefits that would not have happened without the program. Before approving the budget, a marketing leader needs to name the customer decision a reward could change.
The practitioner who wrote “Should you invest in a loyalty program?” describes how competitive pressure can push that decision aside. A rival introduces a program, and a few weeks later another brand has presentation slides and a program name. Yet the team still needs to decide what customers should do differently, because enrollment alone does not establish use and issuing points does not solve cancellations.
Making that decision starts with defining what the business plans to offer. A loyalty program gives customers a future benefit earned through repeated purchases, commonly through points they collect and redeem, sometimes alongside status levels or members-only benefits. An immediate, uniform member discount without an earning threshold is a price reduction. Either approach needs a commercial purpose: what additional behavior will the spending produce?
To establish that purpose, the practitioner proposes four questions for the funding discussion: what value builds up as customers use the product, who is likely to join, can typical customers earn a useful reward, and what result would justify stopping? The answers connect the proposed benefits to customer behavior and a test of business value. A client’s existing points program shows why teams need that connection before committing to more tiers or benefits.
A fourth-purchase reward missed customers who had bought only once
The practitioner’s client had run a points program for years and was reconsidering its value. Before redesigning the benefits, the practitioner asked how much the program cost and what business value it should create. To answer, the team needed to compare customers’ purchase histories with the point at which they could claim a reward.
The practitioner’s comparison exposed a mismatch: nearly nine in 10 customers had bought once, while a reward required a fourth purchase. The business needed to persuade those customers to return for the first time, but its incentive required several further transactions. So a reward could exist in the program yet offer little help at the customer’s next buying decision.
The points balances showed how far the problem reached. The practitioner found that about seven in 10 points ever issued were too small to spend, so much of the promised benefit remained out of reach under customers’ actual buying patterns. Increasing the eventual reward would still leave someone who had bought once facing a distant threshold. Instead, the practitioner’s diagnosis pointed toward offering a benefit earlier, while the business was still trying to establish repeat purchasing.
Even while those balances sat unused, they created a financial commitment. The practitioner explains that issued points are promises recorded on the business’s books, so it must have cash available if customers complete the actions needed to redeem them. Low redemption still leaves a budget question: what does the business owe, and what behavior does it expect that commitment to produce?
To test a different use of the budget, the practitioner modeled spending the same amount on customers considering their second purchase. The calculation produced roughly seven times what the existing loyalty program was earning at the time. This was a modeled result; the client then changed its approach, making a reward usable on the second transaction while keeping funding unchanged.
The change gave the funding discussion a specific customer group and purchase decision. As the practitioner suggests putting it: “We can build this, and here’s the group it pays off with.” The team must then explain why those customers could respond, even if they differ from the group in the original proposal. For this client, that meant redirecting existing spending before expanding the program.
The original structure could still look successful, however, because it attracted customers who already bought frequently. A fourth-purchase threshold could matter to those buyers while doing little to help most customers return. To understand the program’s apparent performance, the team needed another comparison: how members behaved before they joined.
A project in mind?
Schedule a 30-minute meeting with us.
Senior experts helping you move faster across product, engineering, cloud & AI.
Better-performing members do not prove the program works
The practitioner’s client drew its members from the one in 10 customers who were already frequent buyers. Their purchasing made the membership base look attractive, but the program could not claim credit for behavior that preceded it. Comparing those members with people who had bought once would largely mean comparing customers whose relationships with the business were already different.
Daniel McCarthy, an associate professor of marketing at the University of Maryland, describes the problem directly: “Loyalty programs often get credit for loyalty that was already there,” he told the practitioner. The practitioner illustrates this with a hypothetical comparison six months after enrollment. Members might spend more or remain customers at a higher rate than nonmembers, but those differences alone cannot establish how much joining changed their behavior.
McCarthy identifies two reasons for that limitation: “Members may look better because the customers most inclined to stay are also the most inclined to enroll, and looking only at surviving members compounds that bias.” The first is selection bias: people who join already differ from people who do not. The second is survivor bias: excluding customers who leave makes the remaining group appear more likely to stay.
Blue Apron illustrates how examining survivors can change the picture. In a Knowledge at Wharton interview, McCarthy discussed the company’s public figures showing that six months after signing up, 70% of customers were gone. Those left were already the customers most likely to remain. His example concerned customer losses and the characteristics of those who stayed; it was separate from a Blue Apron loyalty-program evaluation.
For the practitioner’s client, these effects explained how strong member performance could coexist with rewards aimed at a small part of the customer base. Frequent buyers remained valuable customers whose relationships the business might want to strengthen. But their earlier loyalty did not show what the program added, which raised a further question: what already gave them a reason to return?
First ask what already gives customers a reason to stay
To find that existing reason, the practitioner suggests comparing a customer of two years with someone who arrived yesterday. The useful difference is something the longer-standing customer would lose by leaving. Saying “they know how it works” is too weak to establish meaningful value built up through use. What has the customer acquired or arranged that makes continuing worthwhile?
Some purchases leave little of that value behind once collection or delivery is complete. For a standalone or disposable purchase, saving a purchase profile can make the second, third, and fourth transactions quicker and easier. That convenience removes effort, but it gives customers a limited reason to choose the business again. The practitioner’s reasoning leaves a clearer role for an incentive when the previous transaction creates little continuing benefit.
The practitioner’s own move between two cloud-storage platforms shows how a product can create a stronger reason to stay. The files moved successfully, but sharing settings and links already sent to other people still relied on the old account. Rebuilding those arrangements took work. The accumulated setup had given the original platform value beyond storing the files themselves.
That continuing value changes what an additional reward needs to achieve. Where customers have substantial setup to preserve, points may pay them to continue behavior the product already encourages. Where each purchase leaves little behind, the business may need to give customers a fresh reason to choose it each time. A reward’s usefulness depends partly on which situation the customer faces.
Make the reward usable at the next decision that matters
The product’s continuing value helps establish whether rewards have a role, while purchase frequency determines whether customers can reach them. The practitioner argues that points become useful when customers buy often enough to earn and spend them. Designing around an unusually active member can therefore produce a program that looks attractive but offers little practical value to a typical buyer.
The gap becomes especially clear for purchases made once every year or two. At that interval, a customer may forget the seller entirely, which the practitioner reports having done personally. Points sitting in an account have little influence if the customer does not remember the business when another need arises. So the design also needs to account for how the relationship continues between purchases.
Increasing the reward cannot by itself keep that relationship active or bring redemption closer. A modest benefit available at the next purchase may influence that decision more than a larger benefit several transactions away. The client’s second-purchase incentive addressed this timing problem. Other businesses need to find the relevant decision in their own customers’ buying patterns before choosing when a reward becomes available.
For customers with long gaps between purchases, contact may serve a purpose before another transaction is due. The practitioner proposes a $5 Starbucks coffee “reward” once a quarter as a small benefit that could maintain a positive association with the business. This is a suggested relationship reminder, separate from an established Starbucks program. Its purpose would be to keep the business in the customer’s mind between buying occasions.
A reminder and a purchase incentive therefore need different tests. A benefit tied to a transaction asks the customer to buy, while a periodic gift may aim to preserve attention and goodwill until a later need arises. Each needs a clear purpose, a usable benefit, and timing suited to the decision it seeks to influence. Another business’s use of points cannot determine those choices.
Define the comparison and the shutdown condition before launch
Once the team has chosen the behavior and reward, it needs a way to measure the program’s contribution. McCarthy recommends leaving some customers outside the program to form a comparison group that receives no program treatment. Comparing their outcomes with those of participating customers helps separate the program’s effect from behavior that would have occurred anyway. The intended value for customers and the business should determine what gets measured.
The comparison becomes more useful when the team also agrees in advance what result would justify pausing or ending the program. Otherwise, enrollment or continued spending by existing customers can become a standing reason to renew its budget. A clear stopping condition makes that decision part of the original investment case and gives the team a basis for acting when the measured contribution falls short.
The contribution being tested can include protecting an existing customer relationship, even when an immediate additional purchase is not the goal. A business may reasonably spend money to retain valuable frequent buyers. But enrollment, previous loyalty, and stronger results among members do not establish that the rewards provide protection. The comparison needs to test whether those rewards help preserve the relationship beyond the strength it already had.
Key takeaways for leaders
- Define the behavior change: Marketing leaders need to identify the customer decision a reward could influence before approving a loyalty budget. Tie spending to a specific outcome, such as a second purchase or a stronger existing relationship.
- Match rewards to purchase patterns: In one client’s program, nearly nine in 10 customers had bought once, while rewards required a fourth purchase. Comparing purchase histories with redemption thresholds helped redirect the same budget toward the second transaction.
- Account for existing loyalty: Frequent buyers are often the customers most likely to enroll, making member spending an unreliable measure of program impact. Program owners need to account for customers’ behavior before enrollment and include those who leave.
- Assess the product’s staying power: Product and marketing leaders need to identify what customers accumulate through continued use, such as saved settings or established sharing links. That existing value helps determine where an additional reward could influence a decision.
- Make benefits reachable: Design rewards around typical buying frequency and the next meaningful purchase decision. For customers who buy infrequently, evaluate whether periodic contact or a small benefit could maintain the relationship between transactions.
- Set the test and stopping rule: Program owners need an untreated comparison group and agreed conditions for pausing or ending the program. Measure whether rewards generate additional purchases or help retain customers beyond what their existing relationship would predict.
A project in mind?
Schedule a 30-minute meeting with us.
Senior experts helping you move faster across product, engineering, cloud & AI.


