Loyalty: reward and grow.

Remember that it is not about you. It is about your customers, and if they are given what they need, they give back. Cialdini called it the rule of reciprocation, and it is the oldest mechanism in the book. Listen to them, reward them, and keep talking to them, because out of sight is out of heart, and a regular who has not heard from you in a month is a regular who is being courted by someone else.

Reichheld built a career on the other half of the argument: longer-tenured customers buy more, cost less to serve and bring others. Byron Sharp is the useful sceptic, and I keep him on the desk. His research found that loyalty programs rarely change how much people buy: heavy buyers join because they were already heavy buyers. Kim, Singh and Winer's work on the Pareto rule took the air out of the other cliché too: across packaged goods the top fifth of buyers account for closer to half of sales than the 80% everyone quotes. I take both as a warning about what this stage is for. A loyalty program is not a machine for making people buy more. It is the stage where the brand treats the people who already buy differently from the people who do not, which most brands never do, and it is the stage that quietly covers for failures everywhere else.

The step: promoter

The step is called promoter for a reason. The people in this stage recommend you, and the flow that leaves it, Repeat, sends them back to the first step of the funnel, which is what makes it a flywheel. Everything else in the map pushes a person forward; this is the one loop that pushes new people in.

At a large California retailer I designed five tiers on order count: no orders, one order, two to four, five or more under $840 in lifetime value, and five or more above it. Every flow splits by tier, so the same trigger sends a different message depending on where the person is. The tier that matters most is the one-order group, because that is where 64% of first-time buyers stop. But the top tiers are where loyalty is built: a customer with two or more orders who leaves a cart gets free shipping or a perk, never a discount, because the discount teaches the wrong thing to the people who least need it. I also recommended custom RFM settings: the default splits customers into equal thirds, which with 64% of the base on one order would have put a lot of one-time buyers in the middle frequency group. The lowest score covers the bottom 70%, the top score the top 10%, people with four or more orders.

Rewarding regulars without a sale

A hand-dyed yarn brand lives on its regulars, who were 79% of revenue in the six months to July, and they reorder every month. The reward for them is not a discount, which they would learn to wait for. It is three predictable sales a year at most, Memorial Day, Black Friday and Cyber Monday, and in early November, the biggest code of the year goes to the email list only, in a short window a few weeks before Black Friday weekend. The list gets the best deal, first, as a group. Black Friday itself is kept to chosen items. The regulars are rewarded for being on the list, and the brand keeps its full-price months.

Bezos wrote in his first shareholder letter about offering customers something they could not get any other way. For a small brand that is the whole loyalty program: the list gets what the public does not, and the regulars know it. Cialdini's reciprocation does the rest, and it costs the brand nothing it was not already going to give.

The flow: repeat

The repeat flow is the message that brings a regular back for the next order, and it is the flow that is easiest to let run on a guess. At a pharmacy and health brand the reminder fired near day 89 for almost everyone, because the plan said the second order came at day 73. Across 2,477 customers who came back, the median gap was 111 days, and the old reminder matched an observed reorder 3% of the time. The new flow starts from the purchase, times reminders by product category from 18,412 orders, and stops once someone reorders. A repeat flow that fires on the wrong day is noise to a loyal customer; one that fires on the right day is a service, and the difference between the two is one query against the store's own orders.

The objection: our retention is fine, revenue is flat

A kidney-health subscription brand had held revenue roughly flat for months, and paid and email both reported healthy returns. Counted by first-purchase date, new customers had fallen by more than half since early 2024, from about 282 a month to about 90. About 70% of orders were on a subscription, so renewals from the large 2024 cohorts were covering the gap. Loyalty was doing its job so well that it hid the failure of the stages above it. Nobody would have seen it in a revenue chart. The second line, new customers by month, is the one to put next to it, and it is the line this stage is most likely to hide. Sharp's point lands here from the other direction: the regulars were not the growth. They were the floor.

The same brand's failed-payment flow was reaching 37 of 102 subscribers with a declined card, because its check ran a moment before the store wrote the status. A one-off series to the 58 it had missed brought 12 back, with no control group, so I will not claim the lift. Out of sight, out of heart, and sometimes out of the flow entirely.

What I check first

• Customers by order count. One, two to four, five or more. Then what each group gets that the others do not.
• New customers per month, next to revenue. For 24 months, from first-order dates in the store. If one line falls while the other holds, loyalty is masking acquisition.
• The repeat flow's send day against the median reorder gap. From the store's first and second order dates, not the plan.
• What the list gets that the public does not. If the answer is nothing, there is no loyalty program, only a discount schedule.
• Whether the top tier ever sees a discount. If it does, the program is paying the people who would have bought anyway, which is Ehrenberg's finding about promotions applied to your best customers.

If you take one thing from this stage: treat the people who already buy differently from the people who do not, and watch the new-customer line while you do it. Not every delivery goes well, and not every regular stays one. The next stage is what the brand does when someone is unhappy.