Designing Rewards That Actually Change Customer Behaviour

In 2006, two researchers ran an experiment with car wash loyalty cards that quietly rewired how the world thinks about rewards. One group of customers got a card requiring eight stamps to earn a free wash. Another group got a card requiring ten stamps — but with two stamps already filled in, as a "head start." Both cards demanded the same number of future visits. Yet the customers with the head-start card finished faster and returned more often. Nothing about the reward had changed. Only the story of how close they were to it had.

That study, published by Ran Kivetz, Oleg Urminsky and Yuhuang Zheng in the Journal of Marketing Research in 2006, is the cleanest proof of a truth most loyalty teams still haven't absorbed: rewards don't change behaviour because of what they're worth. They change behaviour because of how they're framed. Get the framing wrong and a generous programme sits unused. Get it right and a modest one becomes a habit.

This is the argument I want to make, and I'll defend it hard: most loyalty programmes in the Gulf and beyond fail not because the rewards aren't valuable enough, but because they were built as discount mechanisms and mislabelled as loyalty. A discount changes a transaction. A well-designed reward changes a behaviour. Those are not the same job, and confusing them is the single most expensive mistake in retention economics.

Why do most loyalty programmes fail to change behaviour?

Most loyalty programmes fail because they reward the purchase that was already going to happen, rather than the behaviour the business actually wants more of. A customer buys groceries, earns points, redeems points on the next basket — and nothing about their frequency, basket size, or channel preference has shifted. The programme has simply given away margin on business as usual.

This happens because loyalty programmes get designed by finance and marketing as a subsidy, not by anyone thinking in behavioural terms about what triggers a habit loop. The result is a points ledger that functions exactly like a discount: customers wait for it, expect it, and feel short-changed the moment it's withdrawn. That's loss aversion working against the business instead of for it — Daniel Kahneman and Amos Tversky's finding, from their 1979 prospect theory work that later earned Kahneman the 2002 Nobel Memorial Prize in Economic Sciences, that losses are felt roughly twice as intensely as equivalent gains. Once a discount becomes the expected price, removing it doesn't feel neutral. It feels like a penalty.

A reward that only lowers price teaches customers to wait for a lower price. That's the whole failure mode in one sentence, and it explains why so many "loyalty" schemes are really just deferred discounting with a nicer name.

What separates a discount from a reward, behaviourally speaking?

A discount reduces the cost of something a customer was going to do anyway. A behavioural reward creates a new reason to act — it targets a specific, chosen behaviour (visit more often, try a new channel, refer a friend, stay through a renewal window) and reinforces that exact behaviour until it becomes routine. The distinction sounds academic until you see it in a P&L: discounts erode margin on existing demand, while well-targeted rewards generate incremental demand that wouldn't otherwise exist. Three tests separate the two:

• Specificity — a discount applies broadly ("10% off"); a behavioural reward targets a named action ("your third visit this month unlocks X"). If you can't name the behaviour being reinforced, it isn't a reward, it's a markdown.
• Timing — a discount is felt as a price change; a reward is felt as a consequence of something the customer did. The gap between action and reward should be short enough that the customer's brain still connects the two.
• Reversibility — take away a discount and customers feel robbed, because loss aversion has turned the temporary into the expected. Take away a reward tier a customer hasn't earned yet, and there's no loss to feel — only a goal still ahead.

This is why airline status tiers outlast most points-for-cash schemes: status is earned through behaviour, not handed out as a price cut, and losing it feels like losing an identity, not losing a coupon. That's a far stickier form of loyalty than any percentage-off ever built.

How does the goal-gradient effect explain why customers speed up near the finish line?

The goal-gradient effect, the mechanism behind the car-wash study above, holds that motivation intensifies as the perceived distance to a goal shrinks — even when the actual distance hasn't changed. Effort increases, visit frequency increases, and abandonment drops sharply once someone can see the finish line. It doesn't need the customer to want the reward more. It needs them to believe they are closer to it.

This is why a ten-stamp card with two stamps pre-filled outperforms an eight-stamp card requiring the identical number of future visits — and it's why the best-designed apps show a progress bar rather than a running point total. A number climbing from 0 to 5,000 feels abstract. A bar filling from 60% to 80% feels almost finished. The goal-gradient effect is not a trick; it's a faithful readout of how human motivation actually behaves near completion, and any programme that hides progress behind a flat balance is leaving the single cheapest lever in behavioural design untouched.

The practical implication is blunt: give customers a visible, believable head start, and break long-term goals into shorter visible legs. A twelve-month tier renewal feels distant and abstract. Four quarterly checkpoints toward that same tier feel close and achievable — same destination, entirely different psychology.

Why do unpredictable rewards create stronger habits than predictable ones?

Because certainty is easy to ignore, and a little uncertainty is very hard to walk away from. When a reward always follows the same action in the same amount, the brain files it as routine and stops paying attention — this is the same habituation that makes a fixed discount feel like the normal price rather than a gift. Introduce variability — a surprise double-points weekend, an unexpected upgrade, a "you've been selected" moment that isn't tied to a visible threshold — and attention returns, because the brain can no longer predict the outcome perfectly. This isn't a licence to gamify recklessly. It's a case for weaving a small number of well-placed surprises into an otherwise transparent structure, because surprise interacts with a second, equally powerful lever: reciprocity. Robert Cialdini's long-standing work on influence documents how an unearned gift creates a genuine, if mild, social obligation to reciprocate — which in a loyalty context means the next purchase, the referral, the extra minute spent giving feedback. A surprise upgrade that arrives with no strings attached buys more goodwill than the same value handed out as an advertised, expected perk.

The best loyalty architectures use predictable rewards to build the habit and unpredictable ones to protect it from going stale. Predictability builds trust in the system; unpredictability keeps the system interesting enough to keep noticing.

Can status and identity be engineered, not just points?

Yes — and this is where most programmes leave the most value on the table. Points are fungible and forgettable. Status is not. The moment a customer thinks of themselves as "a Gold member" or "one of our top clients," the reward has moved from the wallet to the identity, and identity-linked behaviour is far more durable than transaction-linked behaviour. Two behavioural mechanisms explain why status sticks where points slide off:

• The endowment effect. Richard Thaler's foundational work on mental accounting showed that people value what they already possess more than they'd pay to acquire it fresh. A customer who has earned a tier doesn't want to lose the identity attached to it — which is why status expiry dates, handled well, drive more retention activity than almost any other single mechanic in a loyalty programme.
• The IKEA effect. People place disproportionate value on things they've had a hand in building. A rewards structure that lets customers choose their own perks, name their own milestones, or build a personalised benefits bundle creates ownership a generic tier list never will.

This is also where personalisation earns its keep. A reward that suits one customer archetype can feel irrelevant to another — free-shipping perks mean little to a customer who always collects in-store, and lounge access means nothing to someone who never flies. Building out customer archetypes before designing the reward matrix is what turns a single generic tier list into something each segment actually wants to defend.

How do you design a reward system that actually changes behaviour?

Reward design is a discipline, not a brainstorm. It follows a sequence, and skipping steps is exactly how programmes end up as expensive discounting engines. Here is the sequence that holds up under scrutiny:

1. Name the behaviour, not the transaction. Decide precisely what you want more of — more visits per month, adoption of a self-service channel, longer tenure past the churn-risk window — before a single point value is set. If the target behaviour can't be written in one sentence, the reward can't be designed around it.
2. Set the reward close enough to feel earnable, far enough to feel valuable. Use the goal-gradient effect deliberately: a first milestone reachable within one or two natural purchase cycles builds early momentum, with a visible progress marker rather than a flat balance.
3. Separate the discount layer from the reward layer. Everyday price promotions can keep doing their job; the loyalty layer should reward behaviours a promotion can't buy — consistency, advocacy, feedback, channel switching.
4. Build in one unpredictable element. A periodic surprise, unadvertised and unearned by any visible threshold, protects the programme from becoming background noise.
5. Attach status, not just stock. Wherever the economics allow, let the top tier carry recognition that outlives the transaction — early access, named service lines, or small rituals of recognition that turn a tier change into a moment worth telling someone about.
6. Test the loss, not just the gain. Before launch, ask what a customer feels the day a benefit is withdrawn or a tier lapses. If that moment feels like theft rather than an unmet goal, the mechanic needs rebuilding before it does damage at scale.
7. Instrument it as a system, not a promotion. Points ledgers, tier logic, and redemption rules need the same operational rigour as any other core system — which is the practical case for running loyalty through a purpose-built loyalty management platform rather than a spreadsheet bolted onto a CRM.

Every step in that sequence is defensible on its own. Skip step one and you're rewarding transactions you'd have gotten anyway. Skip step six and you're building a loss-aversion trap that will cost you goodwill the day someone's status lapses.

What are the most common ways reward design backfires?

Even well-intentioned programmes fall into the same handful of traps repeatedly. Watch for these:

• Rewarding the median, not the margin. Giving the same reward structure to your most loyal 5% and your most indifferent 60% wastes budget on customers who were never at risk of leaving and under-invests in the ones who might.
• Sludge at redemption. Richard Thaler's concept of "sludge" — friction deliberately or carelessly built into a process — shows up constantly at the redemption step: expiry clauses buried in fine print, points that can only be spent on narrow categories, apps that require six taps to redeem a benefit a customer earned honestly. Sludge at redemption quietly cancels out everything the reward design got right upstream.
• Treating points as currency. Points are not currency; they are a story about how close a customer is to something they want. The moment a programme starts pricing points like cash, customers start doing arithmetic instead of feeling progress — and arithmetic kills the goal-gradient effect stone dead.
• No exit ramp for lapsed status. Losing a tier should feel like a solvable problem ("two more visits and you're back"), not a permanent demotion. Programmes that make status loss final turn an emotional lever into a reason to churn entirely.
• Designing for the average customer, not the archetype. A single rewards menu, applied uniformly, ignores that different customer segments are chasing entirely different jobs to be done — and a menu that suits none of them well suits nobody at all.

Harvard Business Review's 2015 research on the emotional drivers of customer value, by Scott Magids, Alan Zorfas and Daniel Leemon in "The New Science of Customer Emotions," made a related point worth borrowing here: customers who feel a strong emotional connection to a brand are consistently more valuable over time than those who are merely satisfied. A reward architecture that only ever moves price, and never touches identity, status or belonging, is optimising for satisfaction while leaving emotional connection — the far more durable asset — entirely on the table.

The reward is a promise, not a discount

Every loyalty programme makes an implicit promise to the customer: keep doing this, and something good keeps happening. The programmes that thrive are the ones that keep that promise legible — visible progress, believable milestones, status worth defending, and just enough surprise to stay interesting. The ones that fail have usually broken the promise quietly, by turning a behavioural mechanism into a pricing lever and hoping no one would notice the difference. Customers notice. They always do — just not consciously, and not until the day the reward stops arriving and the relationship starts to feel like it was never really about them at all. Design for the behaviour you want, not the transaction you already have, and the loyalty tends to take care of itself.

If you're rethinking what your own programme rewards — and whether it's reinforcing habits or just subsidising them — Renascence's customer loyalty strategy work starts exactly where this article does: naming the behaviour before naming the benefit. For a deeper look at the same question from another angle, see our companion piece on designing rewards that change behaviour.