Do loyalty programs actually work?
Published · Updated
Yes, when the reward is certain, progress is easy to see and joining is quick. Studies show customers come back sooner as a reward gets closer, and finish more cards when a card starts with a stamp or two already on it. Programs fail when joining is hard, progress is hidden, or the reward is a prize draw.
The evidence, briefly
How loyalty cards change behaviour is one of the better-studied corners of consumer research:
- Goal-gradient: customers come back sooner as they near the reward. Kivetz, Urminsky and Zheng (2006, Journal of Marketing Research) saw it with a real café reward card. In plain words, someone on stamp 8 tends to come back sooner than someone on stamp 2.
- Endowed progress: a card that starts partly filled gets finished more often. Nunes and Drèze (2006, Journal of Consumer Research) found a 10-stamp car-wash card with 2 stamps already filled in was finished more often than a blank 8-stamp card, for the same 8 washes.
- Keep it certain: our advice is a reward customers know they'll get, like a free coffee, rather than a chance at a bigger prize.
None of this requires believing any vendor's marketing: these are independent academic studies, published in peer-reviewed journals.
So why do so many programs feel dead?
Because they're usually set up against how people behave. A paper card hides progress for the weeks between visits. A program that needs an app download loses people at the download screen. A 20-stamp card puts the reward too far away to pull. Each mistake can be fixed when you set the program up.
A program that follows the evidence looks like this: join at the counter with a quick QR scan, card in Apple Wallet or Google Wallet, progress updated on their phone after every visit, and a certain reward close enough to see (within weeks for a daily or weekly regular, and within 6 to 9 months for a monthly client). Rewardtree is built to work this way: 60 days free, then A$9.99 a month per shop. That's our end-of-year special for businesses that sign up by 31 December 2026: you keep it for every shop, including shops you add later, for as long as you stay. From 1 January 2027, businesses that sign up pay A$19.99 a month or A$199 a year.
We make Rewardtree, so weigh our advice with that in mind; the studies above are independent of us.
More questions, answered#
What is the goal-gradient effect?#
The closer people get to a goal, the harder they work towards it. The psychologist Clark Hull described it in the 1930s, and Kivetz, Urminsky and Zheng (2006) found it with a real café reward card: customers bought coffee more often as the free one got closer. It's the engine inside every stamp card.
What is the endowed-progress effect?#
People given a head start finish more often. In a car-wash study by Nunes and Drèze (2006), customers given a 10-stamp card with 2 stamps already filled in finished it more often than customers given a blank 8-stamp card, even though both needed 8 washes. Progress that shows early matters.
When do loyalty programs NOT work?#
When joining is slow or needs an app, when customers can't see their progress (paper in a drawer), when the reward is uncertain (prize draws, "surprise" perks), or when the reward is a year away. All four are design mistakes, not proof that loyalty doesn't work.
Do customers actually want another loyalty program?#
They want the reward, not the admin. That's the case for wallet cards: no app, no password, and a card that keeps itself up to date on their phone.
