A customer opens a flavour sampler expecting three vials and finds five. The surprise registers, but does it register as a reward worth chasing again? That question sits at the center of how flavour shops design discovery, and the answer is less obvious than the marketing instinct suggests. The instinct says: make the surprise bigger. The behavioural evidence says: surprise is doing work, but it has a ceiling, and past that ceiling it stops functioning as a reward and starts functioning as noise.
What Variable Rewards Actually Do in a Flavour Context
Variable-ratio reinforcement — the schedule where a reward arrives after an unpredictable number of responses — is among the most durable patterns in behavioural psychology. It's the mechanism B.F. Skinner documented in the 1950s when intermittent reinforcement produced response rates far higher and more persistent than consistent reward. The usual shorthand is that unpredictability makes behaviour sticky.
That shorthand travels badly into flavour retail. A customer trying a new e-liquid or concentrate isn't pulling a lever for a pellet; they're sampling something they intend to consume, evaluate, and possibly reorder. The reward isn't the surprise itself — it's the match between expectation and experience. A shop that overdelivers on surprise without overdelivering on match produces a spike of attention followed by nothing.
The 62% figure in the title isn't a measured constant. It's a working heuristic that shows up repeatedly in informal tasting panels and small-n flavour trials: roughly six in ten customers report the surprise element of a sampler as a meaningful part of why they'd try another. The rest report it as neutral or mildly irritating. The gap between 62% and a near-universal 95% is the interesting territory, because it tells you the surprise lever has a natural ceiling that varies by person and by context.
Why the ceiling exists
Surprise is metabolically expensive. The brain allocates attention to prediction errors — the mismatch between what was expected and what arrived — because prediction errors carry information. But that allocation is finite. Once a customer has been surprised twice in a session, the third surprise is no longer informative; it's just more cognitive load. The reward curve flattens and then inverts.
Daniel Kahneman's work on loss aversion adds a second constraint. People weigh losses roughly twice as heavily as equivalent gains. In a flavour context, a surprise that misses — a vial that tastes nothing like its description — registers as a loss, not a neutral. A shop running a high-surprise strategy is therefore running a high-variance strategy, and the variance cuts both directions. The 62% who enjoy the surprise are offset by the customers for whom an unexpected miss is disproportionately costly.
The Difference Between Anticipatory and Consummatory Reward
Behavioural research distinguishes anticipatory reward (the wanting, the pull toward a thing) from consummatory reward (the liking, the pleasure of the thing itself). Flavour discovery lives almost entirely in the anticipatory phase. The customer is drawn by the possibility of a good match — the smell of a new concentrate, the colour of a vial, the name on the label.
This matters because anticipatory reward is easy to inflate and hard to satisfy. A shop can manufacture anticipation with almost no cost: novel packaging, cryptic flavour names, a "you won't guess this one" card in the sampler. But every unit of manufactured anticipation raises the bar the consummatory experience has to clear. If the liquid is merely good, the gap between the pull and the payoff reads as disappointment.
The practical implication is that surprise should be front-loaded into the anticipatory phase and dialled back in the consummatory phase. Let the customer be surprised by what's in the box; don't let them be surprised by what it tastes like relative to the description. The first surprise is free. The second one is a liability.
A concrete reference point
The classic illustration is the 1957 Aronson and Mills study on effort justification, later extended into the broader literature on how people rationalise costly or uncertain choices. Participants who underwent an unpleasant initiation to join a group rated the group more highly than those who didn't — the discomfort was retroactively reframed as evidence of value. The flavour analogue is straightforward: a customer who works slightly harder to identify or prepare a sample tends to rate it more favourably than one handed a finished product. Effort, not surprise, is the underused lever. A sampler that asks the customer to do something — dilute, steep, compare two vials side by side — produces more durable engagement than one that simply withholds information.
Decision-Making Under Uncertainty and the Sampler Design Problem
Flavour choice is a low-stakes, high-frequency decision under uncertainty, which puts it in an unusual psychological category. The stakes are too low for deliberative processing; the frequency is high enough that habits form quickly. This is where the sampler as a format earns its keep — it converts a single uncertain decision into several smaller, correlated ones.
But the format has a failure mode. When a sampler presents five unknowns, the customer's decision rule shifts from "which do I want" to "which do I avoid." Avoidance dominates under uncertainty, and avoidance is governed by loss aversion rather than by anticipated gain. A five-vial sampler with one bad vial can poison the whole set, because the bad one anchors the evaluation. A three-vial sampler with one clearly described "safe" option and two surprises performs better, because the safe option gives the customer a reference point against which the surprises can be judged as gains rather than gambles.
This is the practical case for the 62% framing. The shop that treats surprise as a universal good will over-stuff samplers and watch engagement decay. The shop that treats it as a partial, person-dependent lever will segment: some customers get the fully opaque sampler, some get the annotated one, and the surprise is calibrated rather than maximised.
Competitive Play and the Social Dimension of Trying
There's a third layer that's easy to miss. Flavour discovery is social. Customers compare notes, trade samples, argue about tasting notes the way people argue about anything with a shared vocabulary. This is competitive play in the low-stakes sense — status accrues to the person who identified the obscure note, who called the miss, who found the hidden gem first.
Variable rewards interact with this social layer in a specific way: the surprise that lands becomes a story, and the story is worth more than the liquid. The surprise that misses becomes a warning, and the warning travels faster. A shop designing for discovery is therefore designing for the stories customers will tell, and the asymmetry between good and bad stories — bad ones travel further — should shape how much surprise is loaded into any single try.
The forward-looking move is to build feedback into the sampler itself. A card that asks the customer to rate the surprise, not just the flavour, turns each try into a data point about where that customer's ceiling sits. Over a few rounds, the shop learns which customers want the opaque box and which want the annotated one, and the 62% stops being a population average and becomes a per-customer dial. That's the version of variable rewards worth building toward — not bigger surprises, but better-calibrated ones.