Loyalty programmes rarely fail loudly. They launch, run for a while, produce a number nobody acts on, and are quietly abandoned. The failures below are ordered by how much damage they do, and every one of them is recoverable.
What is the most common loyalty programme mistake?
Letting enrolment decay after launch. Everyone offers the card enthusiastically for a fortnight, then a busy week arrives and it stops being mentioned, and because nothing breaks, nobody notices for months.
The tell is enrolment as a share of transactions rather than as a total. A total only ever goes up, so a programme that stopped working six months ago still shows a growing enrolment count.
What are the mistakes that cost the most?
A handful recur often enough to be worth naming, and the first four account for most of the damage.
| Mistake | What it causes | Fix |
|---|---|---|
| Sign-up the customer must do | Low enrolment; staff stop asking | Staff type the number; card arrives by text |
| No owner for the enrolment rate | Decay within about six weeks | One number, visible weekly, per site |
| Threshold too high | Nobody finishes; the card is ignored | 8 to 10 for frequent trade, 6 for appointments |
| Reward is the cheapest item | Reads as mean; trains cynicism | The customer's usual order |
| Messaging too often | The card gets deleted, permanently | About monthly; fewer if visits are rare |
| Short points expiry | Deletes your seasonal customers' progress | 12 months minimum, 24 if seasonal |
| Tiers at small scale | Labels with no behaviour change | One mechanic, one sentence |
| Measuring totals, not rates | Everything looks fine while it fails | Enrolment rate, completion rate, lapsed list |
| No lapsed-customer list | Churn is silent and unrecoverable | 2 to 3x the normal gap, checked monthly |
| Discounted items earning full credit | Paying twice on the same sale | Decide and state it in one sentence |
| Judging it too early | Abandoned before a cycle completes | Wait one full reward cycle |
Why does the threshold matter so much?
Because an unreachable reward removes the only mechanism the programme has. Progress motivates when the finish line is visible; a card requiring twenty visits from someone who comes monthly is a two-year commitment nobody makes.
The error is usually copying a number from a different trade. Ten visits is generous in a café and absurd in a salon, and the number alone does not reveal which: it only makes sense against how often that business's customers come.
What goes wrong with the reward itself?
Businesses cost it at menu price, panic, and pick the cheapest thing on the list. The customer then receives a filter coffee having bought ten flat whites, and concludes the programme is a trick.
Costed properly the difference is trivial. A €4 drink typically costs under a euro in materials, so the generous version and the mean version differ by a few tens of cents across a whole card, and only one of them makes the customer tell someone.
How does messaging go wrong?
By being frequent and general rather than rare and specific. A wallet card that sends a promotion every fortnight gets deleted, and a deleted card cannot be reached again through any channel.
This makes the wallet channel unusually unforgiving. Email tolerates a bad month because unsubscribing is a deliberate act with friction; deleting a card takes two taps and happens silently.
What measurement mistakes hide a failing programme?
Watching totals instead of rates, and watching averages instead of lists. Total enrolments and total stamps issued both rise whatever happens, and neither names a single customer or implies an action.
- Enrolment rate: enrolments as a share of transactions, not the running total.
- Completion rate: the share of enrolled customers who ever reach the reward.
- The lapsed list: named people whose gap has exceeded two to three times their normal one.
- Redemption: whether earned rewards are claimed, because unclaimed means unmotivated.
What is the mistake businesses make with timing?
Judging the programme before one reward cycle has completed. A café can read a result in five or six weeks; a salon on a six-visit card needs eight to ten months, and a garage needs longer than most businesses will wait.
For the first months the only number worth watching is enrolment rate, because it is the one you can still change. Retention cannot be assessed until a reasonable number of customers have reached the reward at least once.
What is the single highest-return fix?
Remove every action from the customer at sign-up. A staff member types the phone number while handing over the order, and the card arrives by text for the customer to open on their way out.
Enrolment is the only step that happens once per customer and cannot be recovered later. Everything else, the mechanic, the reward, the design, the reporting, can be changed next month. The customers who were never enrolled in the first three months are simply gone.
What mistake do businesses make when the programme is working?
They add to it. A programme that is producing results invites tiers, bonus days, referral bolt-ons and a second mechanic, and each addition costs comprehension that the original scheme had earned.
The test is unchanged: can the whole programme still be said in one sentence at a counter? If a change fails that test, it belongs as a dated promotion announced through the card rather than as a permanent rule.
Is it a mistake to run a loyalty programme at all?
In two situations, yes. Where customers never return, a business serving passing tourists in a location nobody revisits, retention has nothing to work with. And where the underlying product or service is the problem, a loyalty card simply documents people leaving.
The second case is worth taking seriously, because a loyalty programme will surface it. A card that enrols well and never gets completed is telling you something about the business rather than about the programme, and that information is more valuable than the rewards.
How do you recover a programme that has already stalled?
Start with enrolment rate, not with the reward. Stand at the counter for twenty minutes and count how many transactions result in an enrolment; if it is under one in twenty, the flow or the script is the problem and nothing else matters yet.
- Watch a real shift before changing anything: the failure is usually visible in ten minutes.
- Rewrite the counter sentence as a statement, not a question, and teach it verbatim.
- Check the threshold against how often your customers come, not against what a competitor uses.
- Send one message to everyone holding an unfinished card telling them where they stand.
- Only then reconsider the reward, which is the thing owners change first and should change last.