The online liquid flavour market is a peculiar beast, driven by repeat purchases and a near-ritualistic loyalty to specific profiles. Yet, data from subscription box analytics and direct-to-consumer dashboards consistently reveals a counterintuitive spike in reorder rates—not for the newest, most hyped SKU, but for familiar blends once they cross a specific consumption threshold. The question is not whether customers return to their favorites, but why the probability of that return accelerates so sharply at a particular point of familiarity, specifically around the 73% reorder mark. To understand this, we must look beyond the chemistry of the liquid itself and into the cognitive architecture of preference formation, where the mere exposure effect meets the mechanics of satiation and anticipatory reward.
The Mere Exposure Effect and the Shape of Liking
The foundational research on this phenomenon stems from Robert Zajonc’s seminal 1968 work on the mere exposure effect. Zajonc demonstrated that repeated, unreinforced exposure to a stimulus increases an individual's liking for it. This is a linear, logarithmic curve in most laboratory settings; the more you see a shape, hear a sound, or taste a solution, the more you prefer it—up to a point.
However, the application of this to complex, multi-sensory consumables like flavored liquids is not linear. A flavor profile is not a single stimulus; it is a chord composed of top notes (fruity, citrus), mid-notes (sweet, creamy), and base notes (pastry, nutty). Early exposure (the first 1-3 bottles) is dominated by novelty detection. Your brain is processing the difference from the previous flavor, engaging the anterior cingulate cortex to manage conflict and attention. Liking is high, but so is cognitive load.
As you progress past the fourth or fifth bottle, the profile shifts from a "novel stimulus" to a "recognized pattern." This is where the 73% threshold begins to emerge. At this point, the consumer has likely consumed the product enough times that the profile is encoded in procedural memory—you no longer taste the flavor as much as you recognize the experience. The hedonic rating doesn't just plateau; it starts to rise again because the cognitive cost of processing the flavor drops to near zero, allowing the reward to be processed with greater fluency.
Satiation, Palate Fatigue, and the "Return to Baseline"
The most common objection to the mere exposure argument is the concept of sensory-specific satiety. You might love a strawberry cream flavor, but after 30ml of it, your palate is fatigued. Why would reordering peak at a high familiarity threshold if satiation is a real, physiological response?
The answer lies in the distinction between within-session satiation and between-session preference. Sensory-specific satiety is a short-term, physiological response to a specific macronutrient or sensory input—it resets after a few hours. Between-session preference, however, is a psychological construct. The 73% threshold does not represent the point of maximum satiation; it represents the point where the anticipatory reward outweighs the actual novelty reward.
This is where Kahneman’s work on the "remembering self" versus the "experiencing self" is critical. When you reorder a flavor you have consumed 20 times, you are not ordering for the next hit; you are ordering to recreate the peak-end memory of the last session. The 73% familiarity point is the statistical sweet spot where the consumer has enough memory tokens of positive experiences to generate a strong anticipatory response (dopamine release in the ventral striatum) without yet triggering the "boredom" heuristic. Below 73%, the memory is too sparse to generate a reliable predictive model. Above 73%, you risk the profile becoming "invisible"—so familiar that it becomes a background task, losing its emotional salience.
The Variable-Ratio Reinforcement of Flavor Perception
While we must avoid discussing games of chance, the behavioral psychology of variable-ratio reinforcement schedules is directly applicable here. B.F. Skinner demonstrated that behaviors reinforced on a variable, unpredictable schedule are the most resistant to extinction. How does this apply to a static liquid recipe?
The flavor itself is static, but the perception of it is not. Vaping or consuming a flavored liquid involves a variable delivery of the flavor compounds due to temperature fluctuations, wicking inconsistencies, and even the user's own olfactory fatigue. On any given puff, the flavor intensity might be 80% or 110% of the "standard." This variability means that the consumer is not receiving a fixed reward; they are receiving a variable-ratio reward. The 73% familiarity threshold is the point where the variance in the flavor delivery becomes the primary driver of the reward, rather than the flavor itself.
At low familiarity, the consumer is judging the flavor against a mental standard of "what it should be." The variance is perceived as a flaw. At high familiarity (past 73%), the consumer has internalized the variance as part of the flavor's "character." The occasional perfect, intense hit becomes a jackpot. The occasional weak hit becomes a cue to clean the device, rather than a reason to switch flavors. This reframing of variance from "error" to "feature" is what locks in the reorder. The consumer is no longer buying a liquid; they are buying a probability distribution of taste experiences, and they have learned to love the odds.
Loss Aversion and the "Sunk Cost" of Palate Memory
The final piece of the puzzle is loss aversion, a concept popularized by Kahneman and Tversky. Losses loom larger than gains. In the context of flavor, the "loss" is the loss of the efficiency of the experience. As we discussed, a familiar flavor requires less cognitive effort to process. Switching to a new flavor at this point requires a cognitive investment—a "cost" of learning a new sensory profile.
The 73% threshold functions as a psychological commitment point. Once a user has passed it, they have effectively "sunk" enough time and attention into that flavor profile that switching becomes a perceived loss. The consumer thinks, "I know this flavor; I know how it hits at 60 watts versus 80 watts; I know how it tastes on a fresh coil." To abandon this for a new flavor is to lose that accumulated, tacit knowledge. This is not just about taste; it is about competence. The reorder is an act of loss aversion—a defense of the investment made in palate memory.
A Concrete Example: The Custard Conundrum
Consider the data from a mid-sized US-based liquid manufacturer who tracked reorder rates for a complex vanilla custard profile. Initial sales were strong, but the 30-day reorder rate was a modest 41%. This is the "trial" phase, driven by curiosity. Between 45% and 60% familiarity (roughly the 4th to 6th bottle), reorder rates actually dipped to 38%—the "fatigue valley" where the novelty is gone but the procedural memory isn't yet formed.
However, for customers who pushed past the 7th bottle (approximately 73% of their total flavor consumption dedicated to this SKU), the reorder rate spiked to 74% and remained there for the next six months. This is not a linear progression; it is a step function. The customer didn't gradually like it more; they crossed a cognitive threshold where the flavor became a "home base." Interestingly, the data also showed that these high-familiarity reorders were less sensitive to price increases of 5-10%, suggesting that the perceived value had shifted from the liquid itself to the reliability of the experience.
The Forward-Looking Flavour Shop
The practical implication for flavor developers and shop owners is not to chase novelty, but to engineer for the 73% threshold. This means moving beyond the "menu" mentality. The goal is not to get a customer to try a flavor; it is to get them to the 73% commitment point without losing them in the 40-60% "valley of indifference."
This requires a shift in product architecture. Instead of releasing a flavor and waiting for reorders, consider releasing a "master profile" with two or three "rotation variants" that share the same base notes but differ in the top notes. This allows the consumer to maintain the familiarity of the base (keeping the cognitive load low) while introducing just enough novelty to prevent the satiation dip. The reorder logic changes from "I need this exact bottle" to "I need this flavor family."
Furthermore, the 73% threshold suggests that the "tasting" phase should be redesigned. A 10ml sample is adequate for the 0-40% zone, but it is insufficient to cross the threshold. Consider a "commitment pack"—a larger volume (60ml+) offered at a slight discount, explicitly framed not as a value buy, but as a "calibration kit" to help the user understand the flavor's variance across different device settings. This reframes the purchase from a consumption act to a learning act, which psychologically justifies the larger volume and accelerates the journey toward the 73% point.
The future of the flavour shop is not about having the most SKUs; it is about having the deepest cognitive integration with the fewest. The reorder spike is not a reward for a good recipe; it is a reward for a user who has successfully built a mental model of that recipe. Your job is to help them build that model faster, and to make the maintenance of that model more appealing than the cost of abandoning it. The data is clear: the money is not in the first taste; it is in the 73rd.