A loyalty program is not a relationship. It is a discount, agreed in advance, redeemed at a time you do not control, by customers who have self-selected as the ones most likely to come back.
That is not an argument against having one. It is an argument for pricing it before you launch it. Most stores install a points app because the category seems to expect it, set the earn rate by copying what looks normal, and never work out what a redeemed point costs them or which orders would have happened anyway.
Our position is that a program pays in a narrow set of conditions: people buy often enough for points to accumulate before they are forgotten, the margin can carry the reward without pushing the price up, and the program gives you something a plain email flow cannot. Outside those conditions, the same money spent on a better second-order experience does more, and it does not create a liability sitting on your books.
Purchase frequency decides this before anything else
Points only work when they add up to something within a period the customer can remember. If your product is bought once a year, a points balance is an abstraction. By the time it matters, the customer has forgotten the login, the brand of the app, and often the store.
So start with your own data. Look at the gap between first and second orders for customers who have ordered twice, and look at how many orders a typical returning customer places in a year. If most of your repeat buyers order several times a year, points can reach a useful balance while interest is alive. If your reorder cycle is long, a program mostly adds a step to checkout.
There is a second frequency question that gets skipped: how many of your customers ever order twice at all. If the great majority buy once, the program will be funded by the small group who already come back. You are not creating loyalty there, you are discounting it.
High-frequency, low-consideration products are the natural fit. Coffee, supplements, pet food, cosmetics, printer supplies, anything that runs out. Furniture, mattresses, appliances and one-off gifts are not, whatever the app's marketing page suggests.
Points are a discount you promise now and pay later, so price them like one.
Price the reward out of margin before you set the earn rate
Every point issued is a promise to give away margin later. Work out what that promise costs at your real numbers, not at revenue.
Here is a worked example with invented figures, so put your own in. Say your average order is 60 dollars and your gross margin after product cost and shipping is 40 percent, which is 24 dollars an order. You set earning at 5 percent of spend, so a 60 dollar order issues 3 dollars of future reward. If every point were redeemed, that is 3 dollars off 24, so roughly an eighth of your margin on every order, forever.
Redemption is never total, and this is where stores fool themselves. Unredeemed points look like free marketing, but they are a promise you made to a customer who did not find it worth taking. Points that go unused are not a saving, they are evidence the offer was not compelling.
Then ask the harder question. Of the orders that earned those points, how many would have happened anyway? If your repeat rate is what it was before, the program has changed nothing except your margin. That is the number a program has to beat, and the only way to see it is to compare the same store before and after, over a period long enough to cover a full purchase cycle.
The people who join are usually the people already coming back
Loyalty programs suffer from a selection problem that almost no dashboard shows you. The customers who sign up, check their balance and redeem are your best customers. They were going to order again. The program has taken a slice off orders you had already won.
Meanwhile the customers you actually wanted to move, the ones who bought once and drifted, rarely enrol. A points balance is a weak reason to return to a store you have no strong feeling about, especially when the balance sits below anything worth having.
The app's report will show a large share of revenue coming from members and present that as proof. It is not proof. It is your loyal customers being counted twice, once as loyal and once as members. Ask a different question instead: has the share of customers who order a second time changed since launch, and has margin per order held?
If you want to move once-only buyers, the tools that work on them are a good post-purchase experience, a reason to come back that relates to the product, and a well-timed reminder. Those work on people who never joined anything.
Where a program genuinely earns its keep
There are real cases. The clearest is a replenishable product in a category where the customer has easy alternatives and no strong preference. Points give a reason to default to you when the thing runs out, and the purchase cycle is short enough that the balance stays live.
The second is a store with a wide catalogue and a customer who buys across it. Rewards give a reason to explore beyond the one product that brought them in, which raises the number of categories a customer buys from and makes them harder to lose.
The third is referral, which usually sits inside a loyalty app but is a different mechanic. Paying an existing customer to introduce a new one is acquisition spend, and it should be judged against what a new customer costs you elsewhere, not against a repeat rate.
The fourth is data and permission. A program that gets people to create an account can give you consented contact details, purchase history tied to a person, and a reason to send messages that are not discounts. That is worth something, but it is worth checking that a simpler account prompt or an email flow would not get the same thing for less.
Do the cheaper repeat work first
Before a program, there is a list of things that raise repeat purchases without giving away margin on every order, and most stores have not finished it.
A post-purchase sequence that answers questions and teaches people to use the product. A replenishment reminder timed to when the product actually runs out. Subscribe and save on the items people rebuy, which converts a repeat purchase into a schedule and costs you a small saving only on the orders it creates. A win-back message with a reason attached rather than a coupon. Packaging and a returns process that make the first order pleasant enough to repeat.
These are cheaper, they are measurable one at a time, and they do not create an open-ended liability. They also tell you something useful. If none of them move your repeat rate, a points program is unlikely to fix a product or experience problem underneath.
When we look at a store and see no post-purchase flow, no replenishment message and a loyalty app installed, we say so in the plan. The app is doing the job of the flows, badly, and paying for the privilege out of margin.
How to test one without rebuilding the store
If the conditions look right, you can test the idea before committing to the machinery.
Start with a manual version. Offer the reward to a defined group by email for a defined period, with a clear expiry, and see whether order frequency in that group changes against a comparable group who did not get it. You are looking for a change in how often people buy, not for redemptions. Redemptions only prove people like free things.
Set an expiry from the start. A reward with no deadline is a liability that grows and an offer with no urgency. Set the window to something slightly shorter than your typical reorder gap so it pulls the next order forward rather than sitting idle.
Watch margin per order, not average order value. A program that lifts order size while giving the gain back in rewards has moved money around and called it growth.
And give it a full purchase cycle before you judge it. For a fortnightly consumable, that is a couple of months. For a quarterly one, longer. If after a full cycle the repeat rate looks the same as it did before, you have your answer, and switching the program off is a legitimate result rather than a failure.