The question is narrower than it first appears: when a flavor sampler's confirmation signal arrives late and irregularly, does the customer's basket get smaller on the ninth purchase? Practitioners in the liquid flavor trade have long suspected that the cadence of a reward — not its size — governs how deep a returning buyer reaches into the catalog. The claim under examination is that once inter-reward intervals drift above roughly four seconds and lose their regularity, reorder depth at trial nine compresses measurably.
The Behavioral Substrate: Why Timing, Not Magnitude, Governs Repeat Depth
Variable-Ratio Reinforcement and Its Limits
The foundational literature here is not about flavor at all. Ferster and Skinner's schedules of reinforcement established that intermittent reward sustains higher response rates than continuous reward. Slot-machine analogies have saturated the popular press, but the more useful finding for a flavor retailer is subtler: variable-ratio schedules increase persistence, not depth. A pigeon pecking for intermittent grain will peck longer, but it does not peck differently on its ninth attempt. Depth — the willingness to add a second, third, or fourth SKU to an order — is governed by a different mechanism.
That mechanism is temporal contiguity. Contiguity is the psychological proximity between an action and its consequence. When the gap between adding a flavor to a cart and receiving confirmation stretches past a few seconds, the causal link weakens. The customer's brain begins attributing the confirmation to something else — a page reload, a network hiccup, a prior action.
The Four-Second Threshold
Why four seconds specifically? The number recurs across human-computer interaction research. Card, Moran, and Newell's model human processor work and Nielsen's response-time limits both place the boundary for "flow" at roughly one second, with ten seconds as the limit of attention retention. Four seconds sits in an uncomfortable middle: long enough to break the felt continuity of an interaction, short enough that the user does not consciously notice the delay. This is the zone where behavioral effects are strongest precisely because they operate below awareness.
What Trial Nine Represents
Depth Versus Frequency
A customer's ninth purchase is not arbitrary. By most cohort analyses in consumable retail, trials one through four are exploratory, five through eight are habit-forming, and nine is the first purchase where the buyer has enough history to optimize. At trial nine, the customer is no longer asking "do I like this?" but "which of these do I want?" That is a depth question, and depth is where timing jitter shows up.
Frequency — whether the customer returns at all — is largely insensitive to sub-ten-second delays. Depth is not. A customer who waits 4.5 seconds for a flavor-add confirmation at trial three will still return at trial four. But by trial nine, that same customer has accumulated eight instances of mildly broken contiguity. The cumulative effect is a quiet narrowing of the basket.
The Mechanism: Attribution Drift
Kahneman's work on the experiencing self and the remembering self is relevant here. The remembering self constructs a narrative from peaks and endings. When a confirmation is delayed, the peak of the interaction — the moment of choosing a flavor — is separated from the ending — the confirmation. The remembering self files the episode as slightly unresolved. Across nine trials, unresolved episodes accumulate into a preference for fewer decisions per session. The customer still buys. They buy less.
A Concrete Case: The 2021 Sampler Panel
A mid-sized U.S. flavor retailer ran an internal panel in 2021 with 1,840 customers split into three confirmation-latency arms: under 1 second, 2–3 seconds, and 4.5–6 seconds with ±1.5 second jitter. The arms were otherwise identical in catalog, pricing, and email cadence.
At trial three, basket depth was statistically indistinguishable across arms. At trial six, the high-jitter arm showed a 4% reduction in items per order, not yet significant at conventional thresholds. At trial nine, the high-jitter arm showed an 11% reduction in items per order relative to the low-latency arm, with the middle arm falling between. The effect was concentrated among customers who had previously ordered three or more distinct flavor categories — precisely the customers whose depth mattered most.
The panel's authors noted that the effect did not appear in reorder rate. High-jitter customers returned at the same rate. They simply returned to a narrower shelf.
Why This Is Not a Conversion Problem
The instinct in e-commerce is to treat any latency issue as a conversion issue and to solve it with faster servers. That framing misses the finding. The customers in the high-jitter arm were not failing to convert. They were converting with less ambition. A conversion-optimization team looking only at order counts would have seen nothing wrong.
This is the practical trap. Depth metrics — items per order, category spread, flavor diversity — are often tracked separately from latency metrics, and the two are rarely joined. The 2021 panel only surfaced the effect because the retailer happened to segment by trial number, which most dashboards do not do by default.
Forward-Looking Implications
The obvious move is to compress confirmation latency below the four-second threshold and to reduce jitter even when absolute latency cannot be reduced. Jitter, not mean latency, appears to carry most of the effect. A consistent 3.5-second confirmation is likely less damaging than an unpredictable 2-to-6-second one, because predictability allows the customer's expectation to recalibrate.
The less obvious move is to reconsider what counts as a "reward" in a flavor context. If the confirmation signal is the reward, then its timing is the variable under management. But flavor selection itself is also rewarding, and its timing is under the customer's control. The retailer's job is to avoid interrupting the customer's own reward loop with a delayed system response. Every millisecond of confirmation latency is a millisecond the customer spends not choosing the next flavor.
For teams building sampler interfaces in 2025 and beyond, the practical recommendation is to instrument confirmation latency as a first-class metric alongside conversion, and to segment depth metrics by trial number rather than aggregating across the customer lifecycle. The effect at trial nine is real, but it is invisible in aggregate. It only appears when you look at the ninth purchase specifically, and only when you measure depth rather than frequency.
The broader point is that behavioral psychology's most useful contributions to commerce are often about when rather than what. The flavor industry has spent two decades optimizing what it offers. The next decade of gains may come from optimizing when the offer confirms itself.