Most businesses are obsessed with acquisition: more ads, more opening offers, more new faces through the door. Meanwhile there is a quiet leak on the other side. Customers who came once, had a perfectly good time, and never came back. They did not complain. They did not cancel anything. They simply stopped showing up, and nobody noticed.
Customer loyalty is about closing that leak. It is getting the person who already knows you to come back by choice, spend more over time and eventually bring someone with them. And it is almost always cheaper than chasing strangers.
What loyalty is, and what it is not
Loyalty is not discounting. A discount buys a visit; loyalty builds a habit. The difference comes down to three things:
- Recognising the customer: knowing who they are, when they last came and what they like.
- Rewarding repeat business, not just the first purchase.
- Keeping the relationship alive between visits, when they are not buying anything.
A loyal customer is not the one who takes advantage of your offer. It is the one who, with every option available, picks you.
How to build customer loyalty in six steps
1. Work out who your customer actually is
You cannot build loyalty with a stranger. The first step is capturing who buys from you without adding friction: a QR code that saves a loyalty card to their phone is enough to start recognising them on every visit. Without identification, nothing else is possible.
2. Give them a concrete reason to return
This is where the programme comes in: stamps (buy nine coffees, the tenth is free) or points redeemable against rewards. It is the visible goal that pulls the next visit forward.
3. Pick a format with no friction
The best programme in the world fails if the customer has to carry a paper card they will lose or install an app they do not want. The format that works today is a digital loyalty card in Apple Wallet and Google Wallet: already on the phone, added in seconds, updates itself.
4. Automate the contact
The relationship is maintained between visits, not during the sale. Automated campaigns do that work for you: a welcome message, a birthday treat on the right day, a nudge to anyone who has gone quiet, and special treatment for your best customers.
5. Ask for reviews at the right moment
A happy regular is your best acquisition channel. Ask for the Google review when satisfaction peaks, right after they redeem a reward. More stars today means more new customers tomorrow.
6. Measure and adjust
Track three numbers: how many customers come back a second time, the average gap between visits, and who is cooling off. Those three tell you whether to change anything.
What the research actually says
Almost everything written about loyalty is assertion without a study behind it. Four published results genuinely change decisions about your programme.
A card that starts with free progress gets finished almost twice as often
Nunes and Drèze handed out 300 cards at a car wash (The Endowed Progress Effect, Journal of Consumer Research, 2006). One version needed 8 stamps starting from zero. The other needed 10 but arrived with 2 already stamped. The real effort was identical: 8 washes. Completion was not: 19% against 34%.
The practical consequence is that the question is not "how many stamps" but "how many do I ask for and how many do I give away up front". A welcome reward stops being a nice gesture and becomes a design decision.
Customers accelerate near the finish, not at the start
Kivetz, Urminsky and Zheng (The Goal-Gradient Hypothesis Resurrected, Journal of Marketing Research, 2006) watched a real coffee programme: the closer a customer gets to the reward, the more often they buy. Two things follow.
First, how you word the reminder. "You have 3 stamps" and "2 more to go" carry the same information and do not perform the same. Always count down from the goal. Second, where the danger is. The weak stretch of a card is the beginning, because at the end customers push themselves. Put your effort into the first few visits.
Rewards are redeemed in two peaks, not a gentle decline
Inman and McAlister (Do Coupon Expiration Dates Affect Consumer Behavior?, Journal of Marketing Research, 1994) found a pattern that keeps repeating: a spike when the reward is issued, a long trough, and a second spike right before it expires.
That has an uncomfortable reading. Extending the expiry window because too many rewards go unclaimed moves the second peak further out; it does not harvest it. What harvests it is a reminder before the deadline. A long window and a reminder are the same piece of advice cut in half.
Reminders work, and one way of writing them works better
Karlan, McConnell, Mullainathan and Zinman (Getting to the Top of Mind, Management Science, 2016) ran field experiments across three banks. Three findings transfer directly:
- Messages that name the goal and the reward together are the ones that move. So "2 stamps to go for your free coffee", not "2 stamps to go" and not "you have a free coffee waiting".
- Framing it as a gain or as a loss made no meaningful difference. Do not spend time agonising over that sentence.
- Additional late reminders added nothing. Nagging does not pay, so your reminder sequence needs an end.
An honest caveat: this was measured on savings accounts, not coffee. The mechanism is the same, because attention is limited and people forget what is not in front of them, but the size of the effect does not transfer between sectors.
A widely repeated figure that is misquoted
Search for anything about retention and you will hit this one: "increasing retention by 5% increases profits by 25% to 95%." It appears on hundreds of pages and almost none of them cite it correctly.
The source is Reichheld and Sasser, Zero Defections: Quality Comes to Services, Harvard Business Review, 1990. What the article says is that reducing defections by 5% increased profits by between 25% and 85%, with different figures per sector: 85% for a branch bank network, 50% for an insurance brokerage, 30% for an auto service chain. The 95% is not in the original.
It is also a 1990 figure about financial and subscription services, where customers sign a contract and cancel it. A coffee shop does not work that way: nobody cancels, they just stop coming. That is why we do not use it anywhere, and the same goes for "acquiring a customer costs five times more than retaining one", which circulates with no traceable source and no sector.
How many stamps, in practice
The rule that works is not a fixed number, it is a distance in time: the reward should sit two to three weeks of normal visits away for that customer.
- Coffee shop or bakery (near daily): 10 stamps, roughly two weeks.
- Bar or casual restaurant (once or twice a week): 6 to 8.
- Hair salon or beauty (monthly or less): 4 or 5, or points instead.
- Gym: a stamp card fits badly, because members already pay a fee. There the programme rewards attendance, not spend.
If the reward is two months away, most people quit before reaching it. If it is too close, you are giving product away to people who were coming anyway. Your number lives between those two walls.
What nobody can promise you
- How much more a loyal customer spends. You would need the receipts of customers who are not in the programme, and those exist in no system. Without a control group, any percentage is an estimate in disguise.
- Whether the card causes repeat visits or just accompanies them. Someone willing to install a loyalty card was already more likely to come back.
- How much you will grow. That depends on your product, your street and your competition far more than on any software.
What can be measured, and what you should insist on, is your own trajectory: how many of your customers return a second time, how many days pass between visits, and whether that number is shrinking.
Why retaining pays better than acquiring
Winning someone new costs money in advertising and time in earning trust from zero. Someone who already trusts you comes back without you paying to attract them again, usually spends more per visit, and brings other people.
This is not an argument for stopping acquisition. It is an argument for not letting what you already captured leak away. Acquiring without retaining is filling a bucket with a hole in it, and the hole is silent: nobody complains, nobody cancels, they just stop coming.
One honest caveat, because this is where the sector overreaches. There is no credible universal number for how much cheaper retention is. What there is, and what you can actually check, is your own repeat rate before and after.
Not losing customers: spotting the one going cold
"Not losing customers" is the other half of loyalty, and it is mostly a timing problem. Decide what number of days without a visit counts as a warning for that customer, and act before they have written you off.
- The signal: someone who came weekly has not appeared in a month.
- The action: a message with a real reason to return. A reward that is nearly complete, something new, a "we have missed you" with something behind it.
The important detail is that the threshold is relative. Twenty days of silence from a daily customer is an emergency; the same twenty days from a fortnightly one is nothing at all. A single threshold applied to everyone produces a list of alarms nobody can act on, and alarms nobody can act on teach people to ignore all alarms.
Done by hand this is unworkable. Done with an automated campaign it happens the moment someone crosses their own line.
The most common mistake
Setting the programme up and forgetting it. A stamp card with no communication around it is a dead mechanic. What creates repeat business is the combination of recognising, rewarding, staying in touch and measuring, working as one system. That is the whole difference between a piece of card and an actual loyalty tool.