Loyalty strategy
How to Build Customer Loyalty That Changes Behavior
Loyalty is earned by experience. How to design a program that changes what customers buy.

How to build customer loyalty: make the everyday experience reliably good, recognize customers wherever they shop, and reward the specific behaviors you want more of, then prove with a control group that the rewards changed what people buy. Loyalty is earned by experience. A loyalty program is the amplifier, not the source.
Many retailers start at the wrong end. They launch points, watch sign-ups climb, and a year later the CFO asks what these customers would have bought without the program, and whether the difference was worth the reward cost. This guide works from the experience outward, with a test you can run on your own data.
How to build customer loyalty: start with the experience, not the points
You build customer loyalty by removing the reasons customers leave before you add reasons for them to stay. A reward can tip a close decision; it cannot make up for an empty shelf, a failed delivery or a store that does not know who you are.
Picture a weekly grocery shopper whose online order arrives with three unwanted substitutions and whose Tuesday points have not posted by Friday. She may still be a member. She is not loyal, and the next competitor with a better delivery slot will have her.
So before you touch the reward catalog, list the moments that decide whether a customer comes back: finding the product, paying, receiving it, getting help and returning it. Fix the worst first.
Why is customer experience strategy the foundation of loyalty?
A customer experience strategy is the plan for which moments of the shopping journey you will make reliably excellent, for which customers, and how you will measure them. It is the foundation of loyalty because customers repeat experiences, not programs.
In retail, a workable customer experience strategy usually comes down to four promises, kept on every channel:
- Availability: what shows as in stock is in stock, online and in store.
- Recognition: the customer is the same person on the app, the website, at the register and on the phone, with one history and one balance.
- Consistency: one price and one set of promotions everywhere, so nobody feels tricked for choosing the wrong channel.
- Recovery: when something goes wrong, it is fixed fast by someone who can see the whole story.
Recognition is where loyalty and experience meet: a member unknown at checkout cannot be rewarded. If your channels still run on separate data, our explainer on what omnichannel means for retailers shows why one customer view comes first.
Why do retail loyalty programs fail? Seven reasons and their warning signs
Retail loyalty programs fail for a small number of repeatable reasons, and each one leaves a fingerprint in the data. Check these seven before you redesign anything.
63%of nearly 870 US consumers surveyed by Bain & Company and ROI Rocket said they make buying decisions based on the loyalty programs they belong to, as reported in Harvard Business Review. Shoppers do respond to programs. When yours does not move behavior, the design is usually the reason, not the idea.
| Reason | Warning sign in your data | First fix |
|---|---|---|
| 1. You reward what customers already do | Reward cost rises with member sales while visit frequency and basket size stay flat | Lower the flat earn rate and move budget to offers that ask for a change |
| 2. One offer goes to everyone | Redemptions concentrate in your most frequent shoppers; lapsed members barely respond | Build 4 to 6 behavior segments, each with its own goal and offer |
| 3. The first reward is too far away | Most members have never redeemed and balances keep growing | Make the first reward reachable within a normal month of shopping |
| 4. You cannot recognize members when they shop | A low, flat share of sales tied to a member ID, especially in store | Accept phone number, app or card; track identification rate by store |
| 5. Rewards are slow or invisible | "Missing points" contacts to customer service; little balance checking after purchases | Award points in real time and show the balance on the receipt and app |
| 6. Only spend counts | Members buy the same categories and channels as non-members | Reward a first purchase in a new category or channel |
| 7. Success is measured on vanity metrics | Every campaign is reported as a win; nobody can state incremental margin | Hold out a random slice from every offer and report margin |
Behind several of these is a selection trap: your best customers join first, so members always outspend non-members, whatever the program does.
Reason 3 also has a balance-sheet cost. Points that sit unredeemed are a liability finance has to carry, and our guide on how to improve customer retention explains points liability and breakage in plain English.
TipBefore you add anything new, switch off the offer that costs the most and shows no lift. Removing waste is the fastest margin you will find in a loyalty program.
Find out which of the seven is costing you most. Our loyalty program health check scores your program against these failure modes on your own member data, and shows where Valus Loyalty's real-time rewards and live segments would change the result. Book a loyalty program health check.
Loyalty program strategy: how to build one in seven steps
A loyalty program strategy starts with the behavior you want to change and works backward to the reward, not the other way round. These seven steps apply whether you are launching a program or rebuilding one.
- Pick one or two target behaviors, such as a second visit within 30 days for new members or a first online order for store-only shoppers. Write each as a measurable change.
- Segment by behavior, not demographics. New, lapsing, single-category, single-channel and top customers need different offers. Four to six segments is enough to start.
- Fix identification at every channel. Accept a phone number at the register, link app and web accounts, and measure the share of sales tied to a known member by store and channel.
- Design the reward mix: a modest base earn rate, a first reward reachable within a month, and most of the budget held for targeted offers.
- Reward in real time on the actual basket, by item, category, channel and tier, and show the balance immediately.
- Agree the funding model with finance. Decide what you will spend, what suppliers might co-fund, how points expire and what liability you are prepared to carry.
- Build the control group in from day one. Every offer has a random holdout, and success is incremental margin per member, agreed with finance before launch.
With high visit frequency and thin margins, grocers feel step 4 most. Our page on loyalty for grocery, supermarkets and department stores covers the specifics.
Loyalty program examples worth learning from
The best-known loyalty program examples share one trait: each is built around a clear behavior the company wants more of. Here are three, described from the companies' own published terms, with the design idea worth taking from each.
Starbucks Rewards: earn rates that rise with status
Starbucks announced in January 2026 that its updated program would have three levels, Green, Gold and Reserve, according to its investor announcement. Members earn 1 Star per dollar at Green, 1.2 at Gold and 1.7 at Reserve; Gold is reached at 500 Stars in 12 months and Reserve at 2,500, and Stars at Gold and Reserve do not expire. The lesson: status changes the earn rate and the rules, so frequent customers feel the difference on every order.
Sephora Beauty Insider: tiers set by annual spend
Sephora's Beauty Insider terms describe a free Insider level, VIB status for members who spend at least $350 in a calendar year, and Rouge status at $1,000, as set out in the Beauty Insider terms and conditions. The lesson: a reachable middle tier gives regular customers a goal, while the top tier recognizes the small group that drives a large share of sales.
Tesco Clubcard: points that turn into vouchers
Tesco Clubcard members collect points as they shop, and Tesco converts them into vouchers each quarter, at 1 point to 1 penny, which members can spend at Tesco or exchange with Reward Partners, according to Tesco's Clubcard vouchers help page. Tesco also runs member-only Clubcard Prices on the shelf. The lesson: pair a simple everyday earn with an instant price benefit, so members identify on every trip.
| Program | Main mechanic | Behavior it encourages | Idea to borrow |
|---|---|---|---|
| Starbucks Rewards | Stars per dollar, three status levels | Frequent visits and ordering through the app | Let status change how fast members earn |
| Sephora Beauty Insider | Annual spend tiers | Consolidating category spend with one retailer | Make the middle tier reachable for regular customers |
| Tesco Clubcard | Points to quarterly vouchers plus member prices | Identifying on every grocery trip | Give an instant benefit, not only a future one |
Copying a mechanic without the experience behind it rarely works. Each of these programs sits on a business customers already use often.
Does a tiered loyalty program work?
A tiered loyalty program works when the next tier is within reach for a meaningful share of members and the status gives them something they value beyond points. It fails when the top tier is out of reach for almost everyone, or when tiers only change a number nobody looks at.
| Tiers tend to work when | Tiers tend to fail when |
|---|---|
| Purchase frequency is high enough for members to see progress within a few months | Customers buy once or twice a year, so status is never felt |
| Benefits are experiential: priority service, free delivery, early access, better earn rates | The only benefit is slightly more points |
| Progress is visible in the app and on the receipt | Members cannot tell which tier they are in or how close the next one is |
| Top-tier cost is modeled with finance before launch | Top-tier benefits grow faster than top-tier margin |
A useful rule: if you cannot name the behavior each tier is meant to encourage, you have a status ladder, not a strategy.
The Loyalty Incrementality Test: how to prove your program works
The Loyalty Incrementality Test compares members who received an offer with a randomly chosen group of similar members who did not, over the same period. The difference in their spending, converted to margin and minus the reward cost, is what the offer actually earned.
- Pick the eligible audience first, for example every member who qualifies for a "spend $100, get $10 back" offer.
- Randomly hold out a share of them. Ten percent is a common starting point. Random means random, not "the ones without an email address."
- Run the offer to everyone else for a fixed window, such as four weeks.
- Compare average spend per member for the whole group, not just the people who redeemed. Redeemers always look good because they were already your most engaged shoppers.
- Convert the lift to margin and subtract the reward cost. That is your net incremental result.
Illustrative example: a blanket offer vs a targeted one
The numbers below are illustrative, chosen to show the method, not taken from any real retailer.
| Illustrative | Blanket offer to all members | Targeted at lapsing members |
|---|---|---|
| Members who got the offer (10% held out) | 90,000 | 18,000 |
| Four-week spend per member: offer group vs holdout | $140 vs $133 | $60 vs $40 |
| Incremental sales | $630,000 | $360,000 |
| Incremental margin at 25% | $157,500 | $90,000 |
| Redemptions and reward cost at $10 each | 16,200 = $162,000 | 4,500 = $45,000 |
| Net incremental result | -$4,500 | +$45,000 |
Same reward, opposite outcome. The blanket offer would likely have been reported as a success, because redeemers spent far above average.
Two mistakes undo most tests: comparing members with non-members, which measures who joined rather than what the program did, and a holdout too small to separate a result from noise. Ask an analyst to size the holdout before launch.
A 30/60/90-day plan to build customer loyalty
About 90 days is enough to see the first measurable results if you diagnose before you redesign and prove each change before you scale it.
Days 1 to 30: diagnose and set the baseline
- Pull five numbers: identified share of sales, share of members who ever redeemed, reward cost as a share of member sales, points liability, and visit frequency.
- Score the seven reasons above and rank them by the reward cost each wastes.
- Create a universal holdout and agree one success metric with finance: incremental margin per member.
Days 31 to 60: fix the experience and the mechanics
- Close the biggest experience gap you found, such as identification at the register or points that post late.
- Make the first reward reachable within a normal month and replace the largest blanket offer with segment offers.
- Add one non-spend behavior worth rewarding, such as a first online order.
Days 61 to 90: prove, then scale
- Read the holdout results for every change. Keep what made margin, stop what did not, and scale the winners.
- Review tiers, thresholds and expiry with finance so the liability becomes forecastable.
- Put a one-page scorecard of these numbers in the monthly trading meeting.
Where to start
Building customer loyalty rarely needs a new app. It needs an experience worth repeating, members you recognize every time, rewards that ask for a change in behavior, and a control group that tells the truth.
Knoativ designs loyalty programs, including the reward-funding model, and runs them on Valus Loyalty, the retail loyalty platform for real-time rewards, live segments and one customer profile across channels. See how we approach identifying, engaging and retaining customers and how our Assess, Build and Operate services work. If you are a brand selling through other people's stores, read CPG marketing with receipt scanning.
If your program has members but no movement, book a loyalty program health check. Bring your numbers and we will score your program against the seven failure modes and sketch the first Loyalty Incrementality Test on your own data.
Frequently asked questions
How do you build customer loyalty in retail?
You build customer loyalty in retail by making the core experience reliable first: accurate stock, one price everywhere, recognition on every channel and fast problem resolution. Then add a program that rewards specific behaviors, such as a second visit or a first online order, in real time. Prove each reward works with a randomly selected control group before you scale it.
What is a good loyalty program strategy?
A good loyalty program strategy names one or two behaviors to change, segments members by behavior, recognizes them at every checkout, and keeps most of the reward budget for targeted offers rather than a high flat earn rate. It agrees the funding model and expiry rules with finance, and measures success as incremental margin against a holdout group.
What are good loyalty program examples?
Starbucks Rewards, Sephora Beauty Insider and Tesco Clubcard are widely studied loyalty program examples. Starbucks links earn rates to status levels, Sephora sets tiers by annual spend, and Tesco turns points into quarterly vouchers alongside member prices. Each works because it sits on a business customers already use often, and each rewards a clear behavior.
Is a tiered loyalty program better than a flat one?
A tiered loyalty program is better when customers buy often enough to see progress and the higher tiers offer benefits they value, such as free delivery, priority service or faster earning. For low-frequency categories, a simple program with personal offers often works better, because most members would never reach a second tier and status would never be felt.
How do you know if a loyalty program is working?
A loyalty program is working when members who receive its offers buy more, more often or more broadly than a randomly chosen holdout group that did not, and the extra margin exceeds the reward cost. Comparing members with non-members does not prove anything, because your most frequent shoppers are the first to join any program.


