A customer has money to spend and a reason to come back. Your loyalty program gives them a password reset, a points balance, and a reward they can't use yet.
Loyalty programs lose value-seeking shoppers when participation costs more effort than the benefit deserves. A discount can still help. But the customer also judges the delivery promise, the return process, the time they spend finding an offer, and whether the purchase fits their budget. Every extra condition changes the deal.
A loyalty program is the system that gives members benefits for an ongoing relationship with a brand. Its job is to make returning worthwhile. The useful question is what gets better for the customer after the first purchase, and whether the business can afford to keep delivering it.
What shoppers expect from loyalty programs
Value-seeking describes a shopper checking what they receive for the money, time, and uncertainty involved. Sometimes the cheapest option wins. Sometimes a slightly higher price is worth paying because the customer trusts the fit, can return the item easily, or knows the parcel will arrive before they need it.
Deloitte's January 2026 analysis cites earlier research finding that four in ten Americans show deal-driven, cost-conscious, or trade-down behavior. It also reports that non-price factors can account for up to 40% of perceived brand value. That's a ceiling across the research, not a promise that any retailer can replace 40% of a discount with better service.

Money
Show the saving in ordinary currency. A future perk won't solve an unaffordable basket.
Time
Let shoppers find, apply, and confirm the benefit without carrying a separate project.
Uncertainty
Make delivery, fit, exchanges, and exclusions clear before the customer commits.
Price remains part of the purchase. A rewards experience can't rescue an unaffordable basket. Customers with a strict household budget may need an immediate saving more than a future perk, and the program should let them see that saving without decoding a conversion rate.
The broader opportunity sits in consumer value beyond a lower price. Consider a clothing purchase. A member discount reduces the amount paid. A clear size history and a straightforward exchange reduce the risk of buying the wrong item. They solve different problems. Which deserves investment depends on where customers actually hesitate.
Start with the customer task. Read the reasons people contact support after ordering. Watch someone attempt to use an existing reward. Find the point where they leave to compare another store. Those observations give the program a specific job, such as reducing uncertainty about delivery or making an everyday replenishment order easier to place.
A team that starts with tiers and points often spends the next quarter inventing reasons to use them. Start with a recurring problem and the right benefit becomes easier to defend.
Where the reward gets lost
Deloitte's loyalty survey covered 5,564 US adult members in September and October 2025. Respondents enrolled in eight programs on average but actively used five. Forty percent sometimes forgot to redeem. Several behavioral findings concern each respondent's favorite program, so they shouldn't be generalized to every membership. These are survey responses, not experimental proof of additional sales.
Those limits matter. An enrollment count tells you that someone accepted an invitation. It says little about whether the reward is visible when they need it, whether it can be used on the item they want, or whether they can finish checkout after selecting it.
Imagine a household buying a familiar product every six weeks. The program offers a benefit after the fourth order, requires activation before checkout, and restricts redemption to a later month. The household has to keep a small project alive for half a year. A saved basket or usable credit at the next order would fit the actual rhythm much better. This is an illustrative design comparison, not a measured result.
Audit the full trip from earning to use. Include finding the reward, understanding its cash value, checking eligibility, applying it, and confirming that the total changed. Count the steps a customer must remember separately from the clicks required. A one-click redemption still fails if the customer never knew it was available.
Make the relevant benefit visible beside the eligible purchase. Show the saving in ordinary currency where possible, explain exclusions before payment, and confirm the result on the receipt. Preserve choice when automatic redemption would consume credit the customer wants to save. Useful automation reduces work without making that decision for them.
A service team should be able to explain the benefit with the same words the checkout uses. If an exception needs a specialist to interpret it, simplify the rule before writing another tutorial.
Point value also affects the decision. Research by So Yeon Chun and Rebecca Hamilton, summarized by the American Marketing Association, found less redemption with variable point exchange rates than with fixed rates, even when average point value was equal. The studies included retail and hotel contexts. That's evidence about redemption behavior, not a guarantee that a fixed rate increases your profit. A customer may keep waiting because a better conversion looks possible. Show the value available now, and make the choice easy to understand.
Benefits with a reason to exist
REI's membership page lists a one-time $30 fee, free US standard shipping with exclusions, used-gear access, and a one-year return window with exceptions. Its typical 10% annual reward applies to eligible purchases and isn't guaranteed. The package combines savings with benefits connected to owning and using outdoor gear.
My reading of that design is that membership becomes easier to understand when the benefits fit the product's life. The customer can see why delivery, returns, maintenance, or resale would matter. That doesn't establish a retention lift for REI, and copying the benefits into an unrelated store would miss the reason they belong together.
For a replenishment brand, useful recognition might mean remembering the last basket and letting the customer edit quantities before reordering. For footwear, it might mean making an exchange easy to start with the original purchase already attached. For a retailer selling gifts, it might mean a clear delivery cutoff and a receipt that protects the surprise. These are proposed benefits, not claims about programs already operating.
Choose the service promise carefully. Priority support creates expectations that staffing must meet. Free delivery has a cost that depends on order size and destination. Longer return windows can affect resale value and inventory planning. The benefit needs an owner, an operating budget, and a fallback when the promise can't be met.

Basic fairness should stay available to everyone. A member lane can offer additional convenience, but an ordinary customer still deserves accurate delivery information and a workable way to resolve a mistake. Making the baseline deliberately painful turns membership into a fee for avoiding the brand's own problems.
Ask what a customer would miss if the benefit disappeared. A clear answer identifies a useful service. A vague answer about feeling special usually means the team has more design work to do.

Tiers have a carrying cost
Starbucks offers a useful current example of the choices involved. Its Rewards terms describe Green, Gold, and Reserve levels. Gold requires 500 Stars within the relevant twelve-month period. Reserve requires 2,500 new Stars after reaching Gold. Status qualification and redeemable reward balances are tracked separately.
For Green members, qualifying monthly activity can extend the life of Stars that would otherwise expire. Gold and Reserve Stars don't expire while the member maintains that status. These are documented rules as reviewed in September 2026, not evidence that the redesign improved retention.

My concern with tier design is the amount a customer must understand before deciding whether a purchase helps them. Qualification balances, redeemable balances, activity windows, and exclusions create different questions. A program can have sound commercial reasons for each rule and still become difficult to explain at the counter.
A tier should change a meaningful outcome. If every level delivers essentially the same coupons with a different badge, the administrative effort deserves scrutiny. If the upper tier offers a costly service, the business needs to know who qualifies, how often they use it, and what happens when capacity runs out.
Keep the promise stable enough for customers to plan around it. When changing a threshold or benefit, show the old and new rule clearly and explain how existing balances are treated. Review the email, app, checkout, receipt, and support script together. Conflicting explanations create extra contacts and make the reward feel uncertain.
Personalization adds a similar obligation. Remembering a size or a replenishment preference can help, especially when the customer can correct it. A long list of inferred interests may produce more irrelevant messages. Collect the information a benefit needs, explain its purpose plainly, and make changing a preference easy. A customer who stops using a recommendation shouldn't have to fight the system to start over.

Repeat purchases can still lose money
The commercial test is additional contribution after the program's costs. Contribution means the money left after the variable costs of fulfilling those sales. An increase in member revenue can coexist with a worse result if rewards subsidize orders that would have happened anyway.
Suppose an ecommerce business tests a benefit with two randomly assigned groups of 1,000 eligible customers. During the chosen observation window, 250 control customers make a repeat purchase and 300 customers offered the benefit do so. In this simplified teaching example, each repeat buyer places one order, each order has $80 of pre-reward sales, and each generates $24 of contribution before program costs.
The offer produces 50 additional orders in the example, a five-percentage-point increase in repeat purchase rate. That generates $1,200 of additional contribution before program costs. If a $5 reward is redeemed on all 300 treatment orders, rewards cost $1,500. Another $200 in incremental service and messaging expense leaves a $500 loss relative to control.
Notice where the subsidy goes. The 250 repeat orders observed in control represent the counterfactual baseline. The treatment group gets rewards on its entire 300-order population. Paying only on the 50 additional orders in the calculation would quietly omit most of the expense. The experiment can't tell you which individual treatment buyers would've returned anyway.
The result could improve with a lower-cost benefit, higher contribution per order, or additional later purchases. It could deteriorate with returns, fulfillment subsidies, or customers pulling a planned purchase forward. Every version needs an observation period long enough to capture the category's normal buying cycle. Don't call a short-term lift durable retention before later cohorts mature.
Member-versus-nonmember comparisons also need care. People who already like the brand may be more likely to enroll. Random assignment before the offer gives a cleaner comparison. Where that isn't feasible, compare similar customers and state the remaining uncertainty. Track all eligible assigned customers, including people who ignore the offer or fail redemption, so selective participation doesn't make the outcome look better.
For a live decision, agree on a primary measure before launch, a minimum worthwhile effect, and enough observations to distinguish an effect from ordinary variation. The illustration above doesn't include uncertainty estimates and isn't a sample-size recommendation.
Build a smaller test
Pick one recurring customer problem and test one benefit. A retailer with frequent replenishment might choose saved-basket ordering. A seller with exchange complaints might choose a clearer member exchange path. Resist bundling either with a new points currency, a tier launch, and a promotional campaign. Multiple simultaneous changes make the result harder to interpret.
Keep the standard experience acceptable in both groups. Record who was assigned the offer and when, whether they could use it, the cost of each use, and subsequent orders and returns. Agree with support and operations on how to handle failures before customers encounter them.

Use a short measurement plan with a decision attached to every signal:
- Repeat purchase rate: Did more eligible customers return within a category-appropriate window?
- Net contribution per eligible customer: Did additional sales cover rewards, delivery subsidies, returns, and service costs?
- Redemption completion: Could customers use the benefit after trying, and where did failures occur?
- Repeat timing: Did the offer create a later habit or mostly move an existing purchase forward?
- Support contacts: Did the promise remove work or send more customers looking for help?
Separate diagnostic measures from the primary commercial outcome. A higher redemption rate helps explain usage. It doesn't prove profitability. A lower contact rate can reflect fewer problems, but it can also reflect customers giving up. Read a sample of failed attempts and service conversations alongside the numbers.
Set a stopping rule for an operational failure, such as a benefit checkout can't reliably honor. Keep customers who encountered that failure in the experiment's reporting. Excluding them would hide the cost of the actual experience. If the test changes partway through, mark the change and analyze the versions separately.
The team's next move should follow the evidence. Expand a benefit that produces a useful customer outcome at a sustainable cost. Fix a service failure before expanding access. Retire an offer that only generates subsidized purchases, even when its redemption chart looks impressive. Convenience becomes a brand promise when the operation can deliver it consistently.
Make returning worth the trouble
Some customers will decline membership. Let them buy. Others will use the program only when a benefit fits their circumstances. Treat that as information about the offer rather than a reason to send more reminders.
A fair price and reliable fulfillment give the program something worth reinforcing. When the product disappoints or delivery misses a promise, recovery comes first. A points multiplier is a poor response to an unresolved service problem.
The hard decision is often what to remove. Cut the reward nobody can explain. Retire the message that repeatedly reaches people with no eligible purchase. Stop asking customers to reconfirm information the business already has permission to use. Keep the benefit that resolves a specific problem, even if it looks modest in a launch presentation.
At the next purchase, the customer will compare the total effort of buying from you with the alternatives. Give them a reason that survives that comparison. The clearest evidence of loyalty is a customer returning after the offer expires, because buying from you still makes sense.

