The number of samples a customer can take before committing to a bottle looks like a trivial merchandising detail. Yet in a small but growing set of flavour retail experiments, capping the sample allowance at three rather than five has coincided with a roughly 12% lift in repeat purchases over the following ninety days. Why would a smaller allowance produce more loyalty? The question sits at the intersection of behavioural psychology, decision-making under uncertainty, and the reward structures that govern how people learn to like what they buy.
The Sample Counter as a Decision Environment
A liquid flavour shop is, functionally, a laboratory for repeated choice under incomplete information. Every bottle is a gamble on a sensory outcome the customer cannot fully predict from the label. Nicotine strength, base ratio, cooling agent, sweetener level, and steep time all interact in ways that even experienced vapers find hard to model mentally. Sampling is the mechanism that reduces that uncertainty before purchase.
Classical economics would predict that more free information is strictly better. Five samples dominate three. The customer learns more, makes a better-informed choice, and the shop earns goodwill. The 12% repeat-buy figure complicates that prediction, and the complication is where the psychology gets interesting.
Information Has a Cost Side
Herbert Simon's work on satisficing established that decision-makers stop searching once an option clears an internal threshold — they do not exhaustively optimise. Barry Schwartz later popularised the darker implication: beyond a certain point, additional options degrade satisfaction and increase regret. The sample counter is a compressed version of this. Each additional sample is another comparison the customer must hold in working memory, another axis on which the eventual purchase can feel wrong.
At three samples, the comparison set is small enough to rank. At five, the customer is managing ten pairwise comparisons and, in practice, often defers the decision entirely — walking out with nothing or defaulting to the cheapest bottle rather than the best one.
Variable-Ratio Reinforcement and the Flavour Discovery Loop
B.F. Skinner's variable-ratio schedules produce the most persistent response rates of any reinforcement pattern he studied. The key property is unpredictability of reward per attempt, not magnitude of reward. Flavour sampling maps onto this cleanly: some samples delight, some disappoint, and the customer cannot know which before trying.
The behavioural risk is that a five-sample allowance turns the counter into a near-continuous reinforcement schedule. With five attempts, the probability that at least one sample lands well approaches certainty. The customer leaves satisfied but not invested. With three, the hit rate is lower and the resulting purchase carries more of the customer's own judgment — and judgment, once exercised, tends to be defended.
The Endowment Effect on a Self-Made Choice
Once a buyer has committed to a specific bottle after a constrained search, the choice acquires a small dose of self-authorship. Richard Thaler's endowment effect research shows that people value objects more highly once they own them, and a related literature on choice justification shows that people rate chosen options more favourably than rejected ones — even when the choice was arbitrary. A three-sample cap makes the final selection feel more like a decision and less like a coin flip among near-equivalents.
What the 12% Actually Measures
The repeat-buy lift is not a claim that three samples are universally optimal. It is a claim about a specific population: adult customers in a category with high sensory variance, moderate price points (typically $15–$28 per bottle), and no strong brand prior. The mechanism appears to be a shift in which bottle gets bought, not merely how many.
A useful reference point is the jam study conducted by Sheena Iyengar and Mark Lepper in 2000. A display of 24 jams attracted more browsing than a display of six, but the six-jam display converted to purchase at roughly ten times the rate. The lesson generalises: breadth of exposure and depth of commitment pull in opposite directions. Sample counters are the flavour equivalent of a jam table, and the optimal width is narrower than intuition suggests.
Three Samples, Three Functions
In practice, a three-sample cap tends to distribute across distinct sensory roles:
- Sample one establishes a baseline against the customer's current product.
- Sample two tests the most plausible alternative — usually a different nicotine strength or base ratio.
- Sample three resolves a specific uncertainty, often cooling intensity or sweetness.
Five samples rarely produce five distinct functions. They produce redundancy: a second fruit profile when the first already answered the question, a third menthol when two established the range. The marginal sample adds noise rather than signal.
Risk, Regret, and the Cost of Over-Searching
Daniel Kahneman and Amos Tversky's work on loss aversion showed that losses loom roughly twice as large as equivalent gains. In a sampling context, the "loss" is the regret of a bad bottle — money spent, coil fouled, week of unpleasant vaping. A customer who has tried five options has a larger imagined space of better alternatives they declined. That counterfactual inventory is precisely what drives post-purchase regret and suppresses repurchase.
The three-sample customer has a smaller rejected set. Their chosen bottle is not obviously inferior to anything they passed over, because they passed over less. The regret surface is flatter, and flat regret surfaces correlate with repeat behaviour.
Practical Implications for Shop Operators
Forward-looking shops are treating the sample cap as a tunable parameter rather than a fixed courtesy. Three appears to be a useful default for categories with high sensory variance and moderate price points. Categories with lower variance — single-note menthols, for instance — may tolerate four or five without the same dilution effect. Categories with very high variance, such as complex dessert profiles, may benefit from a cap as low as two, paired with a staff-guided narrowing step before the customer reaches the counter.
The larger point is that sample allowances are a behavioural design choice, not a customer-service metric. Measuring repeat purchase rate at 30, 60, and 90 days against allowance tiers is straightforward for any shop with a loyalty system, and the resulting curve is likely to be non-monotonic. The interesting question is not whether to cap samples, but where the cap sits for a given product mix and customer base — and whether that number drifts as the category matures.