What brings a customer back, and how sure anyone can be.
Seven levers for a café, a restaurant or a shop. For each one: what the published evidence says, what to do at your counter, and where the evidence stops. Free to read and to download. We do not ask for your email.
- levers
- 7levers
- sources, all named
- 8sources, all named
- limits, stated
- 7limits, stated
Last updated November 10, 2026
None of this is a Perkaria result. It is published research and law, read with a small counter in mind. Where a study is about car washes or credit cards, we say so, because it changes how far you should trust it.
Lever 1
Start the card partly filled
What the evidence says
At a car wash, researchers handed out two kinds of loyalty card. One needed eight purchases. The other needed ten, with two stamps already on it. Both groups had exactly eight washes left to buy. The study reports that 34% of the head-start group finished their card, against 19% of the other group. People work harder at something they have already begun than at something they have yet to start.
What to do
If you run a stamp card, consider a longer card that starts with a stamp or two already on it, rather than a shorter empty one. The number of purchases you ask for stays the same; only the starting line moves. Say why the stamps are there, even with a simple reason such as a welcome gift.
Where it stops
This was one car wash, with a few hundred cards. It measured how many people finished a card, not whether the business earned more. And it only works as described when the head start is honest: a card that needs more purchases than before is a price rise, and customers notice.
Sources : [1]
Lever 2
Show them how close they are
What the evidence says
In a café rewards program, members bought coffee more often the closer they came to a free one. In the same research, people given a twelve-stamp card with two stamps already on it filled it faster than people given an empty ten-stamp card. The authors also observed that the pace dropped back once the reward was claimed.
What to do
Make progress visible at every visit, and say it out loud: two more and the next one is on us. If you are able to contact your members, the moment they are one or two purchases short is the moment a reminder has a reason to exist.
Where it stops
Buying faster near the reward is not the same as buying more over a year. Part of the effect is timing: visits that would have happened anyway, brought forward. The slowdown after each reward is in the same study, and it is the part most people leave out.
Sources : [2]
Lever 3
Expect a modest effect, and keep the program cheap
What the evidence says
A widely cited review of loyalty programs concluded that they seldom build loyalty beyond what the product already earns. They mostly reward purchases that would have happened anyway. Their main benefit is defensive: they help an established business hold its ground, at the price of higher marketing spending.
What to do
Fix the product, the welcome and the wait before you fix the program. Then run a program that costs you little per visit, so a modest effect is still worth having. Judge it on visits from members over several months, not on how many people signed up in the first week.
Where it stops
That review dates from 1997 and looked mostly at large programs such as airlines and national retailers. A counter where the owner knows the regulars by name is a different setting. Whether its conclusion holds for you is something only your own numbers can show.
Sources : [3]
Lever 4
Do your own arithmetic
What the evidence says
You will be told that keeping 5% more customers raises profit by 25% to 85%. The figure is real: it comes from a 1990 Harvard Business Review article. It is also about large service companies whose customers stay for years and are worth more each year they stay.
What to do
Ignore the famous number and use four of your own: your average ticket, your margin on it, how often a regular comes, and what the reward costs you to hand over. A program pays when the extra visits it brings are worth more than the rewards and the fee. If you cannot make that sum come out with honest inputs, a vendor's statistic will not rescue it.
Where it stops
The article never studied a café, a bakery or a salon. Quoting its range for a neighbourhood counter stretches it well past what its authors examined. We quote it here only so you recognise it when you see it.
Sources : [4]
Lever 5
Never pay for a review
What the evidence says
Google's policy for content on Maps lists two things a business must not do. One is to offer incentives, such as payment, discounts, free goods or services, in exchange for posting any review. The other is to discourage negative reviews or selectively solicit positive ones.
What to do
Ask every customer the same way, with the same link, and offer nothing for it. Do not screen people with a private rating first and send only the happy ones to Google. Ask at a moment that makes sense, such as after a few visits, and ask once.
Where it stops
This is a platform's policy, not a law, and Google can change it. Read the current text before you build anything on it. Canadian rules on misleading advertising may also apply to reviews; that is a question for a lawyer, not for a guide.
Sources : [5]
Lever 6
Ask before you message
What the evidence says
Canada's anti-spam legislation requires consent before you send a commercial message by email or text. The message has to identify who is sending it and give a way to unsubscribe. In Québec, Law 25 adds obligations on how you collect, keep and explain your use of personal information.
What to do
Collect consent with a box the customer ticks themselves, never one ticked in advance, and keep a record of when they agreed and to what. Put an opt-out in every promotional message and honour it at once. Collect only what you will use: a first name and a number go a long way.
Where it stops
This is a summary, not legal advice, and both laws have details and exceptions that depend on your situation. The Canadian Radio-television and Telecommunications Commission and the Commission d'accès à l'information publish their own guidance. For a decision that matters, ask a lawyer.
Sources : [6] [7]
Lever 7
Treat a gift card as a debt that does not expire
What the evidence says
In Québec, the Consumer Protection Act treats a gift card as a prepaid card. As a general rule it cannot carry an expiry date, and the merchant cannot charge fees that eat into the balance over time. The money a customer loaded stays owed until it is spent.
What to do
Sell gift cards, because they bring cash in before the visit. Then keep a running total of what is outstanding and treat it as money you owe, not as money you earned. Make it easy for a holder to check a balance without calling you.
Where it stops
The Act has exceptions, and how to record unredeemed balances in your books is an accounting question. The Office de la protection du consommateur explains the rule in plain language. Your accountant is the right person for the bookkeeping.
Sources : [8]
What this guide is not
- A promise. Nothing here tells you how many more visits you will get.
- A study of our own. We have one deployment, and we do not turn it into statistics.
- Legal or accounting advice. Two levers describe laws in general terms and point you to the people who can advise you.
- Complete. Seven levers is what we could support with a named source and state the limits of. We left out the claims we could not trace.
Sources
- Nunes, J. C., & Drèze, X. (2006). The Endowed Progress Effect: How Artificial Advancement Increases Effort. Journal of Consumer Research, 32(4), 504–512.
- Kivetz, R., Urminsky, O., & Zheng, Y. (2006). The Goal-Gradient Hypothesis Resurrected: Purchase Acceleration, Illusionary Goal Progress, and Customer Retention. Journal of Marketing Research, 43(1), 39–58.
- Dowling, G. R., & Uncles, M. (1997). Do Customer Loyalty Programs Really Work? MIT Sloan Management Review, Summer 1997.
- Reichheld, F. F., & Sasser, W. E., Jr. (1990). Zero Defections: Quality Comes to Services. Harvard Business Review, September–October 1990.
- Google. Maps User Generated Content Policy, section on rating manipulation. Consulted on November 10, 2026.
- Canada's anti-spam legislation (S.C. 2010, c. 23).
- Act respecting the protection of personal information in the private sector (CQLR, c. P-39.1), as amended by Law 25 (S.Q. 2021, c. 25).
- Consumer Protection Act (CQLR, c. P-40.1), provisions on prepaid cards.
Each reference was checked against its publisher's page on the date this guide was last updated, except where the publisher's page could not be opened, in which case it is cited from the reference itself. Read the originals before you rely on any of them.
Want the next guide?
The guide above is yours either way. If you would like the next one when it comes out, this is the only list we keep.
Ready to meet your regulars?
Book a 20-minute demo — we'll set up a sample program with your brand on it, live, while we talk.