The first reorder is the most informative event in a customer's life with a flavour line. It is also the most fragile. A shop that hands a new buyer twelve samples is optimizing for the moment of purchase and quietly taxing the moment of return. Why would a wider sample set, which by any intuitive logic should raise the odds that someone finds something they like, depress the rate at which they come back for a full bottle?
The Paradox of Plenty in Flavour Sampling
The classic demonstration here is Iyengar and Lepper's 2000 field study at a gourmet grocery in Menlo Park. A tasting table offered either six or twenty-four varieties of jam. The larger display drew more foot traffic — roughly 60 percent of passing shoppers stopped, versus 40 percent at the smaller table — but the purchase rates inverted sharply: about 30 percent of the six-jam tasters bought a jar, against roughly 3 percent of the twenty-four-jam tasters. More options attracted more attention and converted far less of it.
Flavour sampling carries an extra cost that jam does not. A jam buyer can evaluate a single spoonful and know. A flavour buyer is evaluating a fit — with a device, a habitual strength, a preferred sweetness level, a throat hit they expect and rarely articulate. Twelve samples do not merely multiply choices; they multiply the number of dimensions on which a buyer can get it wrong. The result is what Barry Schwartz later popularized as the paradox of choice, though the mechanism is better described in terms of decision cost: each additional option raises the cognitive price of committing, and commitment is precisely what a reorder requires.
What Twelve Samples Do to the Memory of the First Purchase
There is a second, subtler problem. Sampling is not just a decision; it is an encoding event. What a buyer remembers about a shop is disproportionately shaped by the end of the experience and by its most emotionally charged moment — the peak-end rule identified by Kahneman, Fredrickson, and colleagues in the early 1990s, and the broader finding that remembered utility diverges systematically from experienced utility. A five-sample flight has a clean end: the buyer finishes, has a favourite, and leaves with a resolved feeling. A twelve-sample flight usually ends in fatigue. The last two or three samples are tasted through a fog of palate overload, and the buyer walks out with a muddled set of impressions rather than a ranking.
That matters for reorders because the reorder decision is made from memory, not from the counter. If the remembered experience is "there were a lot of them and I wasn't sure," the buyer's default is inaction. If the remembered experience is "the third one was clearly the one," the buyer has a specific, nameable target to return to. The five-sample flight manufactures that target; the twelve-sample flight disperses it.
The comparison anchor problem
Twelve samples also invite a comparison process that five do not. With a small set, buyers tend to evaluate each option on its own merits — is this pleasant, is this close to what I want? With a large set, they shift toward relative judgment: which of these is best? Relative judgment is fine for a one-time purchase and corrosive for a reorder, because "best of twelve" is a much higher bar than "good enough to buy again." A flavour that would comfortably earn a reorder after a five-sample flight can lose to a marginally superior competitor in a twelve-sample flight and never be bought at all. The shop has, in effect, taught the customer to hold out.
Variable Rewards and the Risk of a Slot-Machine Counter
It is worth being precise about where the behavioural-psychology analogy actually holds, because it is often misapplied. Variable-ratio reinforcement — the schedule where a reward arrives after an unpredictable number of responses, documented by Ferster and Skinner in the 1950s — produces persistent, high-rate behaviour. A sample counter that sometimes yields a revelatory flavour and sometimes yields duds is, structurally, a variable-reward environment. That is not inherently bad; it is why tasting is fun.
The failure mode is when the shop becomes the variable. If a buyer's twelve-sample experience is genuinely random — one hit, eleven misses — the lesson learned is that the counter is a lottery and the odds are poor. Five samples, curated so that three or four are plausible fits, converts the schedule from variable-ratio to something closer to fixed-ratio with a high hit rate. The buyer learns that the counter is reliable, which is the belief that underwrites a reorder. The distinction is not moral but mechanical: unpredictable quality trains persistence in sampling; predictable quality trains persistence in buying.
Loss Aversion and the Cost of a Wrong Full Bottle
Kahneman and Tversky's work on loss aversion — losses loom larger than equivalent gains — explains why the sample-to-bottle transition is so brittle. A full bottle is a substantially larger commitment than a sample, and a wrong bottle is a felt loss: money spent, shelf space occupied, the small indignity of having chosen badly. Buyers anticipating that loss become conservative. They return to the flavour they are most certain about, and certainty is a function of how cleanly the sampling experience resolved.
Here is the concrete case. A shop testing a five-sample versus twelve-sample onboarding ran the twelve-sample version first and saw strong same-day engagement but a first-reorder rate in the low teens. After switching to a five-sample flight — same total inventory, same flavour catalogue, just a curated subset presented at the counter — same-day engagement dipped slightly while first reorders roughly doubled within thirty days. The mechanism was visible in the follow-up conversations: five-sample buyers could name their favourite immediately and often named a second choice as a backup. Twelve-sample buyers frequently could not name a favourite at all, only a general impression that several had been "pretty good."
Designing the five
The practical implication is not "fewer flavours" but "fewer simultaneous flavours." A five-sample flight should be built around a contrast structure — two close variants, two distant ones, one wildcard — so the buyer learns something about their own preference rather than merely ranking. The wildcard matters: it produces the occasional revelation that makes sampling feel worthwhile, and it does so within a set small enough that the revelation is memorable rather than lost in the shuffle.
The forward-looking question for anyone running a flavour counter is not how to widen the sample set but how to make the first reorder the natural next step. That means treating the sample flight as a memory-design problem, not a menu problem: fewer options, a cleaner ending, a nameable favourite, and a backup choice in the buyer's pocket. The shop that sends a customer home with one clear answer has done more for its repeat rate than the shop that sent them home with twelve open questions.