Retention is a maths problem
Acquisition buys you a transaction. Retention buys you a slope, and slopes compound. This is why two shops with identical footfall, identical average baskets and identical margins can drift into completely different businesses within eighteen months on the strength of a few percentage points of repeat rate. One is refilling a leaking bucket every month at full cost; the other is filling a bucket that mostly holds. From the outside, on any given Saturday, they look the same. The uncomfortable implication is that retention is not the reward for good acquisition — it is the entire point of acquisition. A funnel that never loops is not a funnel; it is a very expensive introduction service. Yet retention consistently loses the budget argument, because acquisition is legible and immediate while retention is diffuse and slow. You can attribute a campaign. It is much harder to attribute the absence of a departure. So the money goes where the attribution is clean, which is a good description of how a business can optimise itself into steadily rising costs and flat revenue while every individual decision looks defensible.
The second visit is the hardest
Almost every retention programme is aimed at people who already come back. Tiers, points, birthday rewards, VIP nights — these are mechanisms for rewarding a habit that already exists, which is pleasant but rarely decisive. The moment that actually determines your repeat rate is earlier and much lonelier: the gap between the first visit and the second. At that point the shopper has no habit, no history, no accumulated balance, and no particular reason to choose you over the equivalent shop two hundred metres away. Nothing about the first transaction tells you whether a second will happen — the receipt looks identical whether they return next week or never think about you again. This is also the measurement trap. If you study loyalty among your loyal customers, you will conclude your loyalty scheme is working, because you have selected for the outcome you wanted to observe. The population that matters is the one that came once. They are harder to find, they are not in your VIP list, and they are where your growth is currently leaking away unmeasured.
Designing the return trigger
A return trigger has three parts, and getting two right is the same as getting none. Timing: too early and it is noise arriving before the memory of the visit has settled; too late and the habit has already formed somewhere else. Relevance: generic contact reads as spam no matter how elegantly it is timed, and reads as spam precisely because it is — it demonstrates that you know nothing about the person you are contacting. And a reason that survives scrutiny: something the recipient would agree is a fair thing to say to them. The triggers that work reference something true. The thing they looked at and did not buy. The category they always buy. The visit they actually made, on the day they actually made it. This is why behavioural data outperforms every demographic segment ever built: relevance is not a clever inference about who someone probably is. It is evidence that you were paying attention to what they actually did. Customers can tell the difference immediately, and they are far more forgiving of a company that noticed than one that guessed.
Loyalty is a habit, not a balance
Points schemes are routinely mistaken for loyalty programmes. They are closer to a discount with a delay attached, and they purchase exactly as much devotion as a discount does: none that survives a better offer across the street. A points balance is a switching cost, and switching costs work right up until the moment someone offers to pay it. Real loyalty is duller and far more durable. It is habit — a store that is easy, familiar, reliably has the thing, and does not waste your time. Nobody defects from that for five percent, because the alternative is not a lower price; it is unfamiliarity, and unfamiliarity has a cost that never appears on a competitor's poster. This is where behavioural data earns its keep. Not by calculating a more sophisticated reward, but by making each visit fractionally better than the last — the size in stock, the queue that did not form, the offer that happened to be for something you actually wanted. Points reward the transaction and are forgotten at the door. Habit rewards the relationship and walks back in on its own.
- A small lift in repeat rate compounds; a one-off sale does not.
- The gap between visit one and visit two is where retention is won or lost.
- Points buy transactions. Relevance buys habits.