The Question Hiding Behind a 16% Lift
A flavour shop that introduces a three-tier skill badge system—novice, adept, and master—reports that customers who earn at least one badge reorder 16% more often by their twelfth trial than customers who don't. The number itself is unremarkable; loyalty mechanics produce lifts like this all the time. What's worth interrogating is why a badge, which confers no discount, no free shipping, and no product, changes anything at all. If the effect is real and durable, it tells us something about how people make repeat decisions under uncertainty—and about what separates a flavour shop from a vending machine.
The answer, I think, lives at the intersection of behavioral psychology, decision-making under uncertainty, reward loops, and the peculiar competitive instincts that emerge when a consumer starts treating a purchase as a practice.
Why an Empty Badge Isn't Empty
Variable-Ratio Reinforcement, Applied Gently
B.F. Skinner's work on schedules of reinforcement established that behavior maintained on a variable-ratio schedule—where a reward arrives after an unpredictable number of responses—is more resistant to extinction than behavior maintained on fixed schedules. This finding has been applied, often cynically, to product design. But there's a version of it that isn't exploitative: the reward isn't a jackpot, it's information about your own palate.
When a flavour shop awards a badge for, say, "five distinct citrus profiles logged," the customer isn't being handed a prize. They're being handed a mirror. The unpredictability isn't in whether they'll win; it's in what they'll discover about their own preferences. That reframing matters because it changes the ethical character of the loop while preserving its motivational pull.
Loss Aversion and the Progress Bar
Kahneman and Tversky's loss aversion—the finding that losses loom roughly twice as large as equivalent gains—has a well-documented corollary in goal pursuit: people are more motivated by not losing accumulated progress than by the prospect of gaining something new. A badge tier that is 80% complete exerts a stronger pull than an identical tier that hasn't been started, even though the objective distance to completion is the same.
This is why three tiers outperform a single badge. One tier is a binary state: you have it or you don't. Three tiers create two intermediate states of partial progress, each of which is psychologically "at risk" of being abandoned. The 16% lift by trial twelve probably isn't driven by the badge itself so much as by the customer's reluctance to let a half-filled tier go to waste.
Uncertainty Is the Product, Not the Obstacle
Flavour Is a Genuinely Uncertain Choice
Here's what distinguishes a flavour shop from most retail: the customer cannot fully evaluate the product before purchase. A shirt can be tried on. A book can be sampled. A bottle of e-liquid at 3mg nicotine in a flavour called "Blackcurrant Anise" is a genuine gamble—not against the house, but against your own prediction of what you'll enjoy.
Research on decision-making under uncertainty (Gigerenzer's work on heuristics, and the broader "explore-exploit" literature in reinforcement learning) suggests that people handle this kind of ambiguity better when they have a framework for interpreting outcomes. A badge system supplies that framework. "You've tried four dessert profiles and preferred the custards" is not a reward; it's a reduction in the uncertainty that makes the next purchase feel risky.
The Competitive Layer
There's a second, subtler mechanism. Once badges exist, flavour selection becomes a domain where competence can be demonstrated. This is competitive play in the sense that Csikszentmihalyi described: a voluntary activity with clear rules, immediate feedback, and a match between challenge and skill. The competition isn't against other customers—there's no leaderboard—but against one's own prior self.
That self-competition is what makes the twelfth trial different from the first. By trial twelve, a customer with badges has a history they can outperform. A customer without badges has only a string of purchases.
A Concrete Illustration
Consider a small shop in the Pacific Northwest that ran exactly this experiment over six months. Three tiers: "Explorer" for trying five distinct flavour families, "Curator" for ten, "Architect" for twenty with at least three repeat purchases in a single family. No discounts attached. No points redeemable for product.
By the twelfth trial, badge-earning customers reordered at a rate 16% higher than the control group. But the more interesting finding was in the composition of those reorders: badge-earners were significantly more likely to reorder within a family they'd previously rated poorly on first trial. The badge system appeared to be encouraging retrial—giving customers a reason to revisit a flavour that had failed the first time, on the theory that their palate, not the product, might have been the variable.
That's a meaningful behavioral shift. It suggests badges aren't just retention mechanics; they're epistemic tools. They change how customers interpret their own negative experiences.
What This Means for the Next Design Iteration
If the mechanism is really about reducing uncertainty and protecting accumulated progress, then the obvious next moves aren't more badges or bigger badges. They're moves that sharpen the information content of each tier.
Three directions seem worth testing. First, tier-specific feedback: instead of a generic "Curator" label, tell the customer what their pattern actually is—"you gravitate toward high-menthol profiles" or "your repeat rate is highest on bakery notes." The badge becomes a hypothesis about the self, which the customer can then test. Second, decay mechanics: not punitive, but gentle—a tier that fades if inactive for 90 days, which leverages loss aversion without creating resentment. Third, cross-customer comparison at the tier level only: knowing that 40% of Curators eventually reach Architect is a social fact that costs nothing and adds a competitive frame without exposing individual data.
The 16% figure will not hold forever. Novelty effects decay, and any mechanic that works by making people feel seen eventually stops feeling novel. But the underlying principle—that uncertain choices become tractable when they're framed as skill development rather than consumption—is durable. The shops that internalize that principle will keep finding new ways to apply it. The ones that just copy the badge system will wonder why their lift disappeared by month eight.