Loyalty Program Strategy: Driving Customer Lifetime Value

Users buy once and don't come back

A fashion e-commerce platform with ~2.6 million monthly active visitors had a retention problem. Users would browse, purchase, and then disappear. There was no mechanism to reward repeat behaviour, no reason for a customer to choose this platform over a competitor for their next purchase, and no emotional connection beyond the transaction.

The data backed this up: 25% of active users were dropping off from making repeat purchases because they didn't feel incentivised. Industry research reinforced the opportunity: 73% of millennials and 48% of Gen Z said their purchase habits are directly influenced by loyalty programs. Adding a loyalty program to an e-commerce platform had been shown to increase average order quantity by up to 319% in comparable case studies.

The platform had no loyalty infrastructure at all. This wasn't an optimisation problem. It was a greenfield opportunity to build a retention engine from scratch.

Working backwards from CLV: not features

Before designing any loyalty mechanics, I started with the business goal: increase customer lifetime value. CLV is a function of two levers: purchase frequency and average order value. The loyalty program needed to move both.

I used the Working Backwards framework to pressure-test the idea:

• Who is the customer? Monthly active users, users with declining purchase frequency, and "promo hunters" who spend heavily during campaigns but go silent between them
• What's the problem? Users don't feel rewarded or motivated to make repeat purchases: there's no gratification loop
• What's the value proposition? Every dollar spent earns points. Points unlock cashback, coupons, and tier-based perks. The more you shop, the more you earn, creating a flywheel effect
• What's the projected impact? 40% uplift in customer lifetime value

Two interlocking systems: Points and Tiers

The loyalty program was designed around two complementary mechanics that reinforce each other:

Points (short-term gratification): Every $1 spent earns 1 loyalty point. Points can be converted into cashback, used to unlock discount coupons, or redeemed for exclusive benefits. This creates an immediate, tangible reward for every transaction: the "I earned something" moment that drives the next purchase.

Tiers (long-term progression): As points accumulate, users progress through membership tiers. Each tier unlocks escalating perks: priority delivery, premium support, exclusive access to sales, and partner benefits outside the platform. Tiers create status and aspiration: the "I'm building toward something" feeling that drives loyalty over months, not just transactions.

The two systems feed each other: points drive individual purchases (frequency), while tiers encourage users to spend more per order to reach the next level (AOV). Together, they attack both levers of CLV simultaneously.

Key design decisions:

• Auto-enrolment: every user is in the program from sign-up, zero friction
• 1:1 multiplier as the base, with tier-based multiplier boosts to incentivise progression
• No fee to join: removing any barrier to participation
• Points visible on every screen (cart, checkout, profile) to maintain awareness

Ship points first, prove the model, then layer tiers

Rather than launching the full program at once, I designed a phased rollout where each phase was validated through A/B testing before proceeding:

• Phase 1: Points: Launch the points system to a treatment group. Measure purchase frequency, AOV, and repeat purchase rate against control. This validates whether the core earning mechanic drives behaviour change.
• Phase 2: Tiers: Layer tier progression on top of points. Measure tier adoption rate, spending velocity as users approach tier thresholds, and overall CLV impact.
• Phase 3: Public rollout: Full launch with both systems active, informed by experiment data from phases 1 and 2.

Success metrics were defined at two levels: primary (CLV uplift, NPS) and operational (points issuance rate, redemption rate, tier distribution, and the ratio of points-driven transactions to organic ones).

A look inside the full case study

Here's a preview of what's covered in the complete breakdown: