A loyalty program either gets members to a reward or it does not. Most of the design conversation is about what the reward should be, and most of the outcome is decided by how far away it is.
Pick one goal
Programs underperform when they try to do several things at once, because the design choices pull against each other.
Choose one for the first sixty days:
More visits from existing customers. Bigger baskets from existing customers. Fewer customers drifting away after two or three visits.
These need different builds. Frequency means counting visits. Basket size means counting spend. Stopping the drift means putting the first reward very close to the start.
Trying to do all three produces a program that does none of them well, and worse, one you cannot explain in a sentence.
Set the threshold from real visit frequency
This is the decision that matters most and it is arithmetic rather than judgement.
Find out how often a regular actually comes. Then set the threshold so they reach the reward in six to eight weeks.
A cafe regular visiting twice a week finishes a ten stamp card in five weeks. That works. A barber's customer visiting monthly takes ten months on the same card. That is not a loyalty program, it is a filing cabinet. A salon client visiting every six weeks needs either a spend based program or a reward at three appointments.
Copying a cafe's ten stamp card into a business with monthly visits is the single most common structural mistake in this category, and it is invisible until you notice that nobody has ever redeemed anything.
Build a journey with an early win
A single distant reward asks a new member to take your word for it for two months. Many will not.
Two rewards work better than one. Something small partway, something worth returning for at the end.
The early reward is doing a specific job: it teaches a new member that the program is real and that you honour it, before they have invested much. Once someone has redeemed once, the second card is much more likely to be completed.
If your software allows it, start members with a stamp or two already filled and raise the threshold to match. Twelve stamps with two given beats ten from zero, even though the customer buys the same number of coffees. Nunes and Drèze demonstrated this with car wash loyalty cards in 2006 and it has held up since.
Write terms that can be said out loud
Vague terms create arguments at the counter, and the arguments happen at your busiest moment.
Reward terms should answer three questions: what exactly the customer gets, what is excluded, and when it can be used.
Write them as a sentence a person can repeat, not as a clause. "Any regular drink, in store, not with another offer" beats a paragraph, because it is what staff will actually say.
On expiry, be careful. Expiring stamps caps an open ended liability, and it is also the most common reason customers get annoyed. If you set one, tie it to inactivity rather than a calendar date, make it at least a year, and warn people before it takes effect.
Make it deliverable at the till
A structure that works on paper and not during a rush is not a working structure.
The test: can a new member of staff on their second shift explain the program and deliver the reward correctly without asking anyone?
If the answer is no, simplify until it is yes. Every exception you add is a decision someone has to make under pressure, and different people will make it differently.
Give staff one sentence rather than a briefing, attach it to a fixed moment in the transaction, and check the signup rate weekly for the first month. Getting staff to mention it naturally covers this properly.
Common structural mistakes
A threshold copied from a different trade. Covered above, and it is the big one.
Rewards that need judgement. If staff have to decide whether something qualifies, some will say yes and some will say no.
No interim reward. New members drop out before they learn the program works.
Rules that change. Staff stop trusting them and start improvising.
Optimising for unredeemed rewards. Large programs count on breakage. At a few hundred customers, an unredeemed reward is a customer who lost interest, not a saving.
Checking the structure is right
Three numbers, once the first cohort has cycled through at about eight weeks.
Redemption rate tells you whether the threshold is reachable. Low redemption almost always means it is too far away.
Visits per member before and after joining tells you whether the structure is changing behaviour rather than just recording it.
Percentage of member spend given back tells you whether it is affordable. It should be near what you designed for.
Adjust the threshold before you touch the reward. Thresholds are precise and reversible. Rewards are blunt and reducing one is remembered.
For the full build sequence, how to create a loyalty program. For the structures themselves, types of loyalty programs.
Helpful next reads: