Flavour profiling—the short questionnaire that asks a customer to rate sweetness, cooling, acidity, and aroma intensity before recommending a concentrate—has a well-documented abandonment problem. Most shops that deploy one see completion rates between 40 and 60 percent, and the drop-off clusters at the same place: the moment between submitting the first screen and receiving the first recommendation. A small experiment run across three online flavour retailers in 2024 suggests that a two-second animated progress bar inserted at exactly that gap raised full profile completion by roughly 12 percent. The question worth pursuing is not whether the bar worked, but why a trivial interface element changes a decision that customers describe as low-stakes and voluntary.
The Completion Problem Is a Motivation Problem, Not a Design Problem
It is tempting to treat flavour-profile abandonment as a form factor issue—too many sliders, clumsy mobile layout, unclear wording. Usability audits do find those things, and fixing them helps. But the 12 percent lift from a progress bar points somewhere else. The bar added no information, removed no friction, and shortened no questionnaire. What it changed was the customer's estimate of how much effort remained and whether that effort would pay off.
Behavioral economics has a name for the underlying mechanism: goal-gradient effects. In a classic 1934 study of rat behavior later extended to human consumers, subjects accelerate as they approach a reward—loyalty-card holders, for instance, buy coffee faster when they are two stamps from a free drink than when they are eight stamps away. The perceived distance to the goal, not the absolute effort, governs persistence. A flavour profile with eight questions feels long when the customer cannot see the end and short when a bar shows three-quarters filled.
There is a second mechanism running alongside it. Uncertainty about what happens after submission—will the recommendation be any good, will it be a sales pitch, will I have to talk to someone—inflates the perceived cost of finishing. A progress bar is a weak but genuine commitment device: it implies a defined endpoint and a defined payload. It answers the unspoken question "what am I actually getting for these next four clicks?"
Variable Reward and the Ethics of the Recommendation Reveal
The recommendation itself is where flavour retail brushes against the psychology of reward loops, and this is the part that deserves more scrutiny than it usually gets. When a customer finishes a profile and receives a suggested concentrate, the quality of that suggestion varies. Sometimes it nails the profile; sometimes it misses. That variability is structurally similar to a variable-ratio reinforcement schedule, the intermittent-reward pattern that produces unusually persistent behavior in both animal and human subjects.
Two things follow. First, variable rewards are why some customers return to retake a profile after a bad match—they remember the good hit, not the average. Second, and more important for anyone running a flavour shop: variability is only defensible when the recommendation is honest. If the profile is a thin pretext for pushing whichever concentrate has the best margin, the reward schedule becomes a manipulation, and customers eventually notice the pattern. The behavioral literature on trust is unambiguous here—violations are weighted more heavily than equivalent positive surprises, a manifestation of loss aversion in the relationship itself. A profile that sometimes misses but is transparently trying to match is a different product from one that reliably recommends the house blend.
The practical implication is that the progress bar and the recommendation engine have to be designed together. Promising a payoff with a countdown and then delivering a generic answer trains the customer to discount the next promise.
Why Two Seconds, and What the Timing Signals
The specific duration in the experiment matters less than what it communicates. A progress bar that completes in under a second reads as decorative; one that takes ten seconds reads as a stalling tactic and invites abandonment mid-animation. Two seconds sits in the range where customers perceive that something is being computed on their behalf—a small, legible piece of work. This is the same reason a brief "matching your profile" state increases satisfaction with recommendation quality even when the underlying algorithm is unchanged.
Kahneman's work on experienced versus remembered utility is relevant. Customers do not remember the aggregate of their interaction; they remember the peak and the ending. A profile that ends with a visible, quick computation and a specific recommendation is remembered more favorably than an identical profile that ends with an instant, unexplained result, even if the recommendation is the same. The two-second bar is, in effect, a manufactured peak at the moment of handoff.
There is a risk of over-reading this. The 12 percent figure comes from a modest sample and has not been replicated at scale; treat it as a directional finding, not a benchmark. Shops that have tried longer or more elaborate animations report diminishing and sometimes negative returns, which is consistent with the idea that the signal is "work is being done," not "entertainment is being provided."
Risk-Taking, Competitive Play, and the Customer Who Enjoys the Gamble
A subset of flavour customers treat profiling as a game rather than a chore. They retake it, compare results across sessions, and enjoy the mild uncertainty of whether a new recommendation will land. This is the same disposition that shows up in competitive play and in consumer risk-taking more broadly: a preference for outcomes that are not fully determined in advance, provided the stakes are low and the feedback is fast.
Shops can serve this segment without crossing into manipulation by making the uncertainty explicit and bounded. Letting a customer see two candidate recommendations and choose, or offering a "surprise me within my profile" option, converts the variable reward from something done to the customer into something the customer opts into. Research on perceived control consistently shows that voluntary uncertainty is experienced as pleasant while imposed uncertainty is experienced as stressful—the same variable schedule produces opposite responses depending on who chose it.
The distinction has commercial weight. Customers who feel they opted into a gamble retake profiles and recommend the shop to friends; customers who feel a gamble was run on them churn quietly. The interface element that raises completion by 12 percent is neutral on its own. What determines whether it builds a durable customer base is whether the reward it promises is real, specific, and chosen.
Where This Goes Next
The productive frontier for flavour profiling is not more elaborate questionnaires or longer animations, but tighter coupling between the promise made during the profile and the specificity of what arrives at the end. A progress bar that signals "a real match is coming" only works while real matches keep coming. The next round of experiments worth running would test whether completion gains hold when the recommendation quality is held constant and the bar is varied, and separately whether customers who receive a genuinely well-matched concentrate after a visible wait show higher repeat-profile rates than those who receive an equally good match instantly. That second question is the one that separates a conversion trick from a product worth returning to.