The conventional wisdom in assortment planning holds that less is more: trim the menu, reduce the cognitive load, and watch conversion rise. But the behavioral evidence on choice overload is messier than the slogan suggests, and it points somewhere counterintuitive for anyone running a liquid flavour shop. The question worth asking is not whether smaller assortments convert better in the abstract, but whether a six-flavour lineup produces more trial reorders — the second purchase of a flavour a customer has already sampled — than a four-flavour lineup does. That distinction matters, because trial reorder is where novelty-seeking behaviour meets reinforcement, and the two forces pull in opposite directions depending on how many options sit in front of the buyer.
The Original Choice Overload Finding and Its Limits
Iyengar and Lepper's 2000 jam study is the canonical citation: shoppers presented with 24 varieties were more likely to stop and sample than shoppers presented with six, but only about 3 percent of the large-assortment group made a purchase, against roughly 30 percent in the small-assortment condition. The paper became shorthand for "fewer options, more sales," and it has been cited in assortment strategy ever since.
What gets lost is the boundary condition. The 24-jam condition was extreme, and subsequent meta-analyses — Scheibehenne, Greifeneder, and Todd's 2010 review of 50-plus experiments — found a mean effect size near zero, with substantial heterogeneity. Choice overload is real, but it is not linear. It appears when options are difficult to compare, when the decision is consequential, and when the chooser lacks clear preferences. It fades when options are easy to differentiate and the cost of a wrong pick is low.
For a liquid flavour shop, both of those moderators cut against the "go minimal" conclusion. Flavour names are highly legible — a customer can read "vanilla custard" and "blue raspberry" and form a rough expectation in under two seconds. And a single bottle is a low-stakes purchase. Under those conditions, the overload penalty shrinks, and the benefit of a wider set — more chances that one option matches an existing preference — grows.
Why Six Beats Four for Trial Reorder Specifically
The metric in question is not first purchase. It is the second purchase of a flavour the customer has already tried. That reframes the psychology entirely, because trial reorder is a reinforcement question, not a selection question.
Variable-ratio reinforcement and the sampling sequence
Skinner's work on variable-ratio schedules established that behaviour reinforced after an unpredictable number of responses is more persistent than behaviour reinforced every time. A four-flavour lineup gives a customer a small sample space: by the third or fourth bottle, they have likely tried most of what is available, and the "what's next" signal goes quiet. A six-flavour lineup extends the sampling sequence by roughly 50 percent, which means the customer encounters a novel flavour — and a fresh reward — two or three purchases later than they otherwise would. The reinforcement schedule stays variable for longer.
This is not an argument for infinite variety. It is an argument that the duration of the exploratory phase is a function of assortment size, and that a slightly larger set keeps the exploratory phase alive through the exact window when reorder habits form.
The comparison cost stays low at six
Kahneman's account of System 1 and System 2 processing is useful here. Four options can be evaluated almost entirely by System 1 — fast, intuitive, low effort. Six options push slightly toward System 2, but not so far that the evaluation becomes aversive. The crossover into effortful, unpleasant comparison tends to appear higher, often cited around 10 to 15 options for unfamiliar categories. Six sits in a productive middle: enough differentiation to feel like a real choice, not enough to trigger the abandonment response that Iyengar and Lepper observed at 24.
There is also a loss-aversion angle. With four flavours, a customer who dislikes one has effectively lost 25 percent of the menu. With six, a single miss costs roughly 17 percent. The perceived downside of experimentation is smaller, which raises the willingness to try the unfamiliar bottle — and trying the unfamiliar bottle is the precondition for the reorder.
A Concrete Illustration
Consider a shop carrying four flavours: two fruit-forward, one dessert, one menthol. A customer buys the dessert option, likes it, reorders it twice, then tries a fruit flavour, likes it, and settles into a two-flavour rotation. The exploratory phase lasted roughly five purchases. Now expand to six by adding a second dessert profile and a beverage-style flavour. The same customer now has two plausible matches to their established preference, plus a novel category to sample. The exploratory phase extends to eight or nine purchases, and the probability that at least one of the new flavours enters the rotation rises with each additional trial.
The mechanism is straightforward: more options increase the probability that a given customer finds a second flavour they like, and customers with two liked flavours reorder more frequently than customers with one. The reorder rate is driven by the size of the liked set, not by the size of the menu per se. Six flavours is a way of increasing the expected size of that liked set without crossing into the territory where comparison becomes work.
What This Means for Assortment Decisions
The practical implication is that assortment size should be tuned to the exploratory window, not minimized for its own sake. A few design principles follow.
First, differentiate clearly. Six flavours that blur together — three variations on strawberry, say — behave like four, because the customer cannot distinguish them and the effective menu is smaller than the stated one. The benefit of six comes from genuine perceptual distance between options.
Second, rotate deliberately. If the six-flavour lineup is static, the exploratory phase still ends. Periodic rotation of one or two slots reintroduces novelty without expanding the menu, which preserves the low comparison cost while extending the reinforcement window.
Third, measure trial reorder separately from first purchase. The two respond to different variables. First purchase is driven by discovery and packaging; trial reorder is driven by the size of the liked set and the persistence of the sampling schedule. A shop that optimizes only for first purchase may rationally shrink the menu, and in doing so damage the metric that actually predicts lifetime value.
Fourth, resist the temptation to keep expanding. The evidence does not support "more is always better." It supports a middle band — wide enough to sustain exploration, narrow enough to keep comparison effortless. Six against four is a testable claim within that band, not a universal law. The right number depends on how distinct the flavours are and how quickly a typical customer cycles through them.
The forward-looking version of this argument is that assortment is a temporal instrument, not a spatial one. The question is not how many bottles sit on the shelf but how many purchases the shelf sustains before the customer stops discovering. Six flavours, chosen for contrast and refreshed on a schedule, keeps that clock running longer than four — and the reorder data will show it.