Why does a five-day reorder rate sit flat for weeks and then double the moment a shop's flavour count crosses fourteen? The pattern shows up in reorder data from small e-liquid retailers often enough that it stops looking like noise. It looks like a threshold, and thresholds in consumer behaviour usually have a mechanism behind them rather than a coincidence.
The Arithmetic of a Full Rotation
Start with the simplest explanation, because it does most of the work. A vaper who buys one bottle every five days consumes roughly six bottles a month. If a shop stocks six flavours, that customer can reorder the same two or three indefinitely without ever feeling the shelf is thin. Variety is technically available; it just isn't necessary.
Push the count past a dozen and something shifts in how the customer plans. Fourteen flavours is not a magic number in any chemical or regulatory sense. It is, roughly, the point at which a buyer can construct a genuinely varied monthly rotation — one flavour per day for two weeks, or a two-week cycle repeated with substitutions — without repeating a single SKU. Below that, repetition is forced. Above it, repetition becomes a choice.
That distinction matters more than it sounds. Forced repetition reads as scarcity. Chosen repetition reads as preference. The same six bottles a month feel different to the person buying them depending on whether they could have bought something else.
Choice Overload Cuts Both Ways
The obvious objection is that more choice should hurt, not help. Barry Schwartz's work on the paradox of choice, and the jam study by Iyengar and Lepper, both show that large assortments can paralyse buyers. Thirty jams on a table produced fewer purchases than six. So why would fourteen flavours produce more reorders than six?
The answer is that the jam study measured a single purchase decision by a stranger. Reordering is a different task: it is a repeat decision made by someone who already has a habit. For that person, assortment functions less as a decision problem and more as an insurance policy. They are not choosing among fourteen options each time. They are choosing whether to stay with the shop at all, and a deep shelf is evidence that staying is safe — the flavour they like will probably still be there, and if it isn't, something adjacent will be.
This is where the threshold behaviour comes from. Below roughly a dozen SKUs, a customer who loses their favourite has no credible substitute and starts looking elsewhere. Above it, substitution is easy, so the relationship survives a stockout. The reorder rate doesn't jump because people are buying more variety. It jumps because fewer people leave.
Variable Reward Without the Slot Machine Framing
There is a second mechanism, and it is worth being careful about, because it is often described badly.
B.F. Skinner's work on variable-ratio reinforcement established that behaviour maintained by unpredictable reward is more persistent than behaviour maintained by predictable reward. A pigeon pecking for a pellet that arrives after an unpredictable number of pecks will peck far longer than one rewarded every time. The finding is robust and it is frequently invoked in contexts where it does not belong.
In a flavour shop, the variability is real but mild. A customer orders a flavour they have not tried. Sometimes it is excellent, sometimes it is merely fine, occasionally it is disappointing. That variance is genuine and it is the reason people keep sampling. What matters for the reorder metric is that sampling only becomes a habit once there is enough inventory to sample without risk — that is, without the fear that a bad pick leaves you with nothing to fall back on.
Below fourteen flavours, a failed experiment is expensive. Above it, a failed experiment costs one bottle out of a rotation that still contains known good options. The reward structure only becomes tolerable when the safety net exists. That is a more defensible reading of the data than the usual "novelty drives engagement" line, and it points somewhere different: the fix is depth, not novelty for its own sake.
Loss Aversion and the Cost of Running Out
Kahneman and Tversky's loss aversion — losses loom roughly twice as large as equivalent gains — explains why the downside of a thin shelf dominates the upside of a curated one.
A shop with eight flavours can argue, correctly, that every SKU is a proven seller. No dead inventory, no waste. But the customer is not evaluating the shop's inventory efficiency. They are evaluating the risk that the shop fails them. Running out of a favourite flavour is a loss. Discovering a new favourite is a gain of similar magnitude. Because losses weigh heavier, the customer's mental ledger tilts toward the shop with more redundancy, even if that shop's average flavour quality is lower.
This is testable in a way that should interest anyone running a shop. If the threshold is real, then the reorder lift at fourteen flavours should be driven disproportionately by customers whose first-choice flavour went out of stock in the preceding period. If those customers reorder at the same rate as everyone else, the loss-aversion story is wrong and the variety-seeking story is doing all the work. That is a clean, cheap experiment: tag stockouts, match them to customer IDs, compare reorder intervals.
What the Threshold Implies for Inventory Strategy
If the fourteen-flavour figure is a real inflection rather than a coincidence of one shop's data, it changes how a small retailer should think about the shelf.
The instinct in a thin-margin business is to prune. Cut the slow movers, concentrate capital in the top ten SKUs, negotiate better terms on volume. That instinct optimises for inventory turns and it may be quietly suppressing the reorder rate that keeps the business alive. The customers lost to a stockout do not appear in any report as a loss. They simply stop ordering, and the shop attributes it to churn.
A more useful frame: the bottom four or five flavours may not need to be profitable on their own. They need to exist so that the top ten can be resilient. Their job is not to sell. Their job is to make the shelf feel deep enough that a single stockout doesn't end the relationship.
That reframes the question from "which flavours earn their shelf space" to "how much redundancy does this customer base need before it stops hedging." Fourteen is where one dataset says the answer sits. Whether it holds at twelve or eighteen for a different customer mix is an empirical question, and the only way to answer it is to track reorder intervals against flavour count over time rather than treating assortment as a static merchandising decision.
The practical move is to instrument the threshold. Log flavour count by week. Log five-day reorder rate by cohort. Watch for the knee in the curve, and when you find it, resist the urge to trim back below it the next time a slow SKU looks like waste. The waste may be the thing holding the rest together.