How to Build a Customer Loyalty Program That Actually Works

If you've ever signed up for a rewards scheme, then forgotten your login, and binned the emails two weeks later, you already know the problem from the customer side. The brand spent time, budget, and probably a fair amount of hairloss launching the thing. You gave them your email, made one purchase you'd have made anyway, and then… poof… nothing happened.

The "actually works" in this article's title is doing a lot of heavy lifting. It's the quiet admission that most loyalty programs are pants. They issue points, email balance summaries nobody reads, and sit in the eCommerce stack as a line item labelled "loyalty" that nobody has the ball bearings to cancel.

This piece is about the other kind. The programs that improve repeat purchase rates, lift CLV, and earn their spot in your stack. Getting there isn't about having cleverer rewards than your competitors. It's about designing the program as a data engine from day one. Let's get into what that means.

Why most loyalty programs don't work (and it's not the rewards)

Ask most marketers why their loyalty program's underperforming and they'll blame the rewards. Not juicy enough. Not branded enough. Not enough tiers. More often though, the real problems lie a little deeper.

The transactional trap. Most programs reward one type of behaviour: buying. To be fair, it is what you’re trying to incentivise, but it's a bit blunt. You spend £1, you get a point. It’s simple enough, but also pretty useless for anything except entrenching the customers who were already coming back. The customer writing reviews, referring mates, opening every email, engaging on Instagram… None of those count. You're paying points to people for doing what they'd do anyway, and ignoring the early signals that would let you lift someone from casual to regular.

The siloed stack. The Shopify loyalty app market has produced a generation of loyalty programs that live entirely inside one external tool. Points go in, rewards come out, and the behavioural data never leaves to do other things. Which means your email platform doesn't know who's in which tier. Your personalisation engine can't surface tier-specific offers. Your CDP has no idea a customer just jumped from bronze to silver. The loyalty program is running in one direction and the rest of your stack is running blind.

The vanity metrics problem. Ask a marketer how their program's doing and you'll usually get something like: "We've got 40,000 members." Brilliant. How many of them bought last quarter? How much more do members spend than non-members? What's happened to churn since launch?

Don’t have the answers? Your loyalty program isn't driving results, but it is costing you money.

Fix those three and you're already ahead of most search-engine-optimized Google guides out there. Most of the competing advice you'll find online addresses the symptoms (better rewards! clearer tiers! nicer emails!) without touching the causes.

Now, I’m not saying fixing those will be easy, so let’s take a look at where to get started.

Three questions before you design anything

Before you pick a platform, map a tier structure, or even decide whether you want points or tiers, answer these. In order. On paper. With actual numbers where possible.

1. What specific behaviour are you trying to reinforce?

"Loyalty" isn't a behaviour. It's an outcome. The behaviour is something like: second purchase within 60 days of the first. Or expansion into a second product category. Or review written after purchase. Or re-order of consumable products at the replenishment point.

Pick one or two specific behaviours your loyalty program should reinforce. If you can't name them, you're not ready to design yet. That’s not a diss if you’re not ready. Just make sure you're not doing a loyalty program for the sake of doing a loyalty program. Do it right from the start.

2. What commercial outcome does that behaviour produce?

A second purchase within 60 days, for most eCommerce brands, is the single biggest predictor of CLV. Reviews boost conversion rate on the product page. Whatever behaviour you're reinforcing, name the commercial consequence, and make sure that it’s worth the cost you're offering to get it.

3. Where does the behavioural data live afterwards?

This is the question that separates programs that compound from programs that peter out. When a customer hits silver tier, does that information flow into your email platform? Your paid media audiences? Your on-site personalisation? Your SMS flows?

If the answer is no, you're running a rewards program, but not a loyalty engine. They're different things, and only one of them builds over time.

So, questions answered. You have a plan… a good plan hopefully. And now it's time to decide how you’re going to get there. 

Pick the loyalty program model that matches your margins

There are about five loyalty program archetypes worth knowing. You don't pick between them by looking at what your competitors do. You pick based on your margin structure, your purchase frequency, and what your customers actually care about.

Points and tiers. The classic. Works well for high-frequency, mid-margin businesses like fashion, beauty, supplements, food and drink. The system is familiar, so your customers understand it, and tiers give you natural milestones for personalisation. However, without clear behavioural triggers behind tier progression, you end up rewarding volume rather than value.

Paid membership. Amazon Prime is the obvious example here. Works if your core value proposition is strong enough that customers will pay for access to it, free shipping, exclusive pricing, member-only inventory. Terrible choice if you're hoping the membership fee itself will produce value. Customers are pretty smart; they can tell the difference.

Surprise and delight. Lower-volume, higher-margin brands (furniture, premium skincare, direct-to-consumer cookware) often get more mileage from unexpected rewards than from visible points systems. The customer doesn't even know there's a program; they just know your brand sends them a lovely note and a free sample three months after they spent £150 on face creams. The trick to this is the targeting, it only works if your CDP knows who to surprise and when to surprise them.

Access and community. Works when your brand has real cultural weight. Early product drops, member-only events, founder Q&As. Mostly for brands whose customers identify with the name on the label. Trying this without that weight will make you look like you're cosplaying. Not ideal.

Hybrid. What most grown-up programs actually are. Points for frequency, tiers for segmentation, occasional surprise rewards for delight, access perks for the top tier. The complexity is real, but so too is the upside.

If you're picking your model based on "which one does everyone else do", stop right there. The right answer depends entirely on your economics, and more than occasionally, your competitors are just wrong.

But the model’s just the bare bones. What decides whether your customers actually use the thing are the mechanical details underneath. 

The details that decide whether anyone actually uses it

Once you've got the model, the details matter more than the branding. A few rules hold up across most categories:

Make the first reward reachable. If a new member can't earn anything interesting in their first thirty days, you've probably lost them. Welcome bonuses, low-threshold first redemptions, and early wins are the difference between a signup and a member.

Reward engagement, not just transactions. Points for reviews, referrals, completed profile data, opted-in channels, time spent on educational content. Every interaction is a chance to learn more about the customer and to keep them active between purchases. The best loyalty programs score engagement nearly as aggressively as they score spend.

Tie tiers to behaviour, not arbitrary spend thresholds. "Spend £500 to reach gold" is a round number that means nothing to the customer. "Buy three times in six months to reach gold" is a behaviour, and it's one your automation engine can actually trigger communications around.

Personalise redemption, not just communication. Most programs personalise the email that says "you've earned points." Very few personalise the rewards on offer. If a customer's bought three moisturisers, showing them a skincare sample as their redemption option converts better than offering them a branded tote bag. This only works if your program is pulling from customer data.

Don't change the rules after launch. You're allowed to add; you're not allowed to devalue without warning. Customers remember when they were promised 100 points for a review and then you quietly moved it to 50. Communicate changes well in advance, or expect a heated Reddit post.

Get all five of those right and you've got a well-designed loyalty program. Whether it compounds or peters out though depends entirely on where it sits in your stack.

Build it into your stack, not next to it

This is the part most articles skip, and it's the most important bit.

A loyalty program attached onto your eCommerce site as a standalone app is, at best, a rewards gimmick. A loyalty program plugged into your customer data is a retention engine. The mechanics look similar from the outside. The results compound in completely different directions.

Here's the flow that makes a program actually work:

Customer behaviour like purchases, engagement, on-site activity, gets captured in one place, the CDP. That data feeds both the loyalty program (points, tier progression) and the rest of the stack (segmentation, automation, personalisation, paid audiences). When a customer moves tiers, every downstream channel knows about it. Email triggers fire. On-site messaging adapts. Web push and SMS flows respond to tier-specific segments. Rewards get personalised based on actual purchase history, not a one-size-fits-all catalogue.

The result is both a better loyalty program and marketing where every channel knows who the customer is and what they've earned. Loyalty members see different homepages, receive different emails, and get different recommendations than non-members because the data's flowing through a single place.

This is what SALESmanago is built for. The Loyalty Program module isn't a separate tool; it's a set of functions inside the same CDP running your email, your workflows, your personalisation, and your segmentation. Which means tier data is native to every other bit of marketing you're doing.

If you're already running your marketing automation through a CDP, building the loyalty program inside it is almost always the right call. If you're not, the question becomes whether a loyalty program is enough of a forcing function to consolidate your stack. Often, it is.

When your loyalty program sits in your CDP it will generate more data than a standalone one ever could. Once you have them connected, the question becomes are actually looking at the right bits.

Measure what actually matters

Here's a quick test. Ask your marketing team what their loyalty program KPIs are. If the answer starts with "members enrolled" or "points issued," the program is being measured like a newsletter.

The metrics that indicate whether a loyalty program is actually working:

• Repeat purchase rate, member vs non-member. The number one question. If members aren't buying more often than non-members, the program isn't reinforcing loyalty. Simple. Get it sorted first.

• CLV delta. What's the lifetime value of a program member compared to a non-member? If the number isn't higher, and by higher, we mean double-digit percentage, the program might not be worth it.

• Redemption rate. If members are earning points and never redeeming them, your rewards aren't compelling enough. Or your communication's failing. Either way, something needs fixing.

• Incremental revenue. What's the revenue from loyalty-driven behaviour that wouldn't have happened without the program? Hardest metric to attribute but an important one.

• Churn delta. Are members churning less than non-members? By how much?

Notice what's not on that list: enrolment numbers, open rates on loyalty emails, app downloads. Those are leading indicators at best and vanity metrics at worst. A program with 50,000 inactive members is worse than one with 15,000 active ones. The bigger number just costs you more to run.

Track those metrics from day one, because you're going to need them sooner than you think.

Launch, then iterate

Even if you follow all the golden advice here, the reality is that V1 of your loyalty program will be wrong. Not catastrophically wrong, but the rewards won't be quite right, the tier thresholds will be off, one of the earn paths won't get used, and something you thought customers would love will land with an almighty shrug.

Plan for it. Build a 90-day review into the launch timeline before you launch. Look at which rewards got redeemed and which didn't. Look at tier progression rates. Look at whether the behaviours you set out to reinforce are actually being reinforced. Adjust.

The one rule when iterating: add more than you take away. Adding a new earn path delights customers. Removing one feels like a bait-and-switch. If you absolutely have to devalue something, communicate it weeks in advance, grandfather existing balances where you can, and expect some noise anyway.

Do that. Launch, measure, adjust, repeat, and after a couple of cycles you'll notice something. The answer to 'how do you build a loyalty program that works' was never really about the loyalty program at all.

The actual answer

The question in the title  “how to build a customer loyalty program that actually works” turns out to have a surprisingly boring answer. Not a cleverer rewards catalogue. Not a slicker tier structure. Not a better-branded app.

The programs that actually work are the ones built on customer data from day one, connected to the rest of the marketing stack from the start, and measured against commercial outcomes that matter to the business. The rewards are the bit the customer sees. The data infrastructure is the bit that makes the difference.

Everything else is discounting with a shiny logo on it.

FAQs about how to build a customer loyalty program