In a 1973 experiment, two psychologists, John Darley and Daniel Batson, sent trainee priests across a campus to deliver a sermon, of all things on the parable of the Good Samaritan. On the way lay a man slumped against a wall, coughing, clearly in distress, in truth a confederate of the researchers. Who stopped to help did not depend on whether the priest had entered the profession for altruistic or for self-serving reasons, that made no difference at all. It depended on whether they had been told two minutes earlier that they were running late. Of those in a hurry ten percent stopped, of those with time to spare sixty-three, a sixfold difference produced by a single piece of information about the state the person was travelling in.
One can dismiss this as charming social psychology, far from one's own business, and that would be exactly the mistake. Anyone who builds customer loyalty programs builds them on the opposite assumption, namely that a person is explained by their stable attributes, by what they are, rather than by the state they happen to be in. The whole infrastructure of a loyalty program, its records, its segments, its trigger logic, answers with great precision the question of who this customer is and leaves untouched the question of what state they arrive in, even though the second question, in the priest study, explained six times as much as the first.
What RFM measures, and what RFM cannot see
The industry's standard tool is called RFM, recency, frequency, monetary, and it is a good tool for the question it answers. It tells you how recently someone bought, how often, and for how much, and from that you build segments that reliably predict who is valuable and who is about to leave. What RFM does not see, and by its very construction cannot see, is the difference between a customer buying out of routine because the purchase has become habit and a customer who, in that same moment, is privately asking whether they still want this relationship. Both look identical in the dashboard. In truth one is more firmly bound than any metric shows, and the other is already half out the door.
A design flaw collects here that is so widespread it passes for the normal state of things, and it is worth giving it a name, because naming is the first condition for seeing it. I call it the Loyalty Misfit: the structural misalignment of a program that optimises for the segment a customer belongs to, while their retention is decided by the state they are in when the message arrives. It is the same family as the Behavioral Misfit we work with at Engaginglab, the gap between what a system rewards and what the situation actually requires, only here it sits in the customer relationship rather than in the organisation.
Two customers who have both just had a good experience behave more alike than two customers of the same segment, one of whom is content and the other on the way out.
How much that state is worth was quantified by a Dutch researcher long before anyone spoke of a Loyalty Misfit. In 2007 Fred Bronner had nearly thirteen hundred people at the University of Amsterdam leaf through a newspaper and then measured how many ads they recalled, broken down by mood. Those who were stressed or unhappy recalled around thirty-five percent of the ads, those who were relaxed or in a good mood reached roughly half again as many. Follow-up studies by the advertising researcher Richard Shotton went further and measured not only recall but liking and price perception: identical offers were rated twenty-six percent more favourably in a good mood and liked considerably more. Same offer, same person, a different state, and suddenly a different product.
Where the autopilot switches off
One might object that a customer's mood cannot be steered, and that is even true, but it is not the decisive point. The decisive point is that there are moments in which a person's state shifts predictably, and that in the loyalty context these moments are not randomly distributed. The psychologist Adam Alter and his colleague Hal Hershfield showed in 2014 that people whose age ends in a nine make big life decisions disproportionately often, because the culturally loaded threshold of a decade briefly lifts them off the autopilot and makes them ask how their life is actually going. First-time marathon runners are forty-eight percent more likely to be nine-enders than chance would allow, an effect that has nothing to do with the number on the ID and everything to do with the moment of reflection it triggers.
Transferred to customer relationships, this means: churn almost never happens at random, but where the purchasing autopilot briefly switches off and the customer asks whether they still want the relationship. I call this moment the Off-Autopilot Moment, and its triggers in the loyalty business are as concrete as the decade threshold for the marathon runner: a contract running out, the round purchase anniversary, the first bad experience, a move, a job change, the birth of a child. That this window is real shows in a number every subscription provider trips over once they measure honestly: according to Brandmovers, forty-four percent of all cancellations fall within the first ninety days, precisely the window in which the evaluation of the relationship is still open and the autopilot has not yet kicked in.
A retention budget spread evenly across all customers wastes most of its effect on people in autopilot mode, where no persuasion is needed because the purchase has long become routine.
The effect belongs in the Off-Autopilot Moment: in the moments when the customer is already asking whether to stay, because that is the only place where reflection happens and therefore the only place where anything can be influenced at all.
The newsletter that finds nobody home
Most loyalty programs send their communication by calendar, not by experience, and that is the Loyalty Misfit in its most everyday form. The message that the customer is valued arrives with the reliability of a quarterly plan precisely when the customer was not thinking about the brand at all, and almost never in the moment when they had just thought the brand was good. A customer who has just had a distinctly good experience is in a state in which the same message lands many times more strongly than it would three days later in the background noise of their routine, and exactly that state evaporates unused, because the system does not read it as a signal but at most as a data point in a sequence.
The reverse holds just as much, and it is the more expensive half. If the most effective communication belongs in the Off-Autopilot Moment, then any message that reaches a customer on autopilot is not merely wasted but mildly counterproductive, because it addresses a relationship the customer was not thinking about and so reminds them of the routine rather than the bond. The moment right after a fulfilled expectation, a resolved complaint, a reached milestone is usually visible in the dashboard but rarely read as what it is, a brief open window that justifies a different intervention than the same customer three days before or after.
What this means for designing a program
This is exactly where the work we do at Engaginglab begins, and the difference from the usual loyalty mechanics lies not in the tools but in the order. The Drive Method does not start with the question of which reward to hand out, but with a diagnosis of the actual behaviour and the state in which it arises. Applied to a customer loyalty program, that means clarifying three things before designing any mechanic: where in a customer's life cycle the Off-Autopilot Moment actually occurs, how to recognise such a break in one's own data, and which communication belongs in a good state rather than in a calendar slot. The onboarding that is almost always placed in the moment of greatest overload, because the first day is by definition the least familiar one, then needs not to be stripped down but to be docked to the state in which the new customer is actually receptive.
This sounds like more work, but to a large extent it is only a reallocation of attention, away from the question of who the customer is and toward the question of when they are reachable, and the fine thing about that shift is that it costs nothing in the planning except thought. Whoever concentrates a retention budget on the Off-Autopilot Moments rather than spreading it evenly buys the same effect for a fraction of the outlay, and whoever couples their communication to the experience rather than the quarter lifts an effect that already sits in the data and that so far nobody has called up. The question, then, is not whether you know more about your customers, but whether what you know describes the right axis.