Annual-billing discounts work for the provider more reliably than they work for the consumer. Laibson's hyperbolic discounting and Samuelson & Zeckhauser's status quo bias explain why.
Halfway through the third month of a new subscription, you'll see a prompt. Maybe it appears as a banner inside the app. Maybe it's an email. The message is always some version of the same offer: switch to annual billing, save twenty percent. The math, presented in two side-by-side columns, makes the choice look obvious. The annual column is smaller. The annual column is what you should pick.
Most people pick it. Most people also, statistically, don't get the savings the math promised. Understanding why takes a brief detour into two well-replicated findings from behavioral economics, both of which the annual-billing prompt exploits with some precision.
The first is what economists call present bias, formalized in a 1997 paper by David Laibson — then a junior faculty member at Harvard, later one of the most-cited economists of his generation — in the Quarterly Journal of Economics. Laibson's argument was that human decision-making weights immediate outcomes more heavily than a standard exponential discount function would predict. A small immediate cost looms larger in cognition than a larger future cost. The two are not commensurate.
Hyperbolic consumers display dynamic inconsistency. They prefer patient long-run choices… But when the moment to act arrives, they choose impatiently. — Laibson (1997), Quarterly Journal of Economics
Apply this to a mid-subscription cancellation decision. The friction of cancelling — the time spent, the cognitive overhead, the small momentary feeling of giving something up — is immediate. The cost of continuing to pay is in the future. Eleven months of $14.99 charges, paid against an annual plan you committed to four months ago, are abstract enough that the brain barely registers them. The cancellation friction wins; the future cost loses; the subscription continues.
The second mechanism is status quo bias, established in a 1988 paper by William Samuelson and Richard Zeckhauser — both at Boston University at the time — in the Journal of Risk and Uncertainty. Their experiments demonstrated that, across an unusually wide range of decision contexts, people stuck with whatever option was framed as the current state, even when an alternative would have been objectively better.
Individuals exhibit a significant status quo bias… A series of decision-making experiments shows that individuals disproportionately stick with the status quo. — Samuelson & Zeckhauser (1988), Journal of Risk and Uncertainty
Annual billing converts what would otherwise be a monthly status-quo decision into a yearly one. The provider's churn rate against a monthly subscriber is on the order of 5 to 8 percent per billing cycle. The same churn rate against an annual subscriber is roughly 1 to 2 percent per month, because the cancellation decision only meaningfully arrives once a year, and even then only briefly before renewal. The math, from the provider's perspective, is straightforward. Annual billing isn't a discount they give you. It's a discount they pay you to absorb a much lower probability of cancellation.
This is not, by itself, a case against annual billing. For services you would actually use across all twelve months, the discount is real, and the math comes out ahead. The case is narrower. Annual billing produces savings only when usage is steady. It produces losses when usage drops in months four through seven — which is the most common usage shape for services people sign up for impulsively, since most subscription usage shows a decay curve, with the highest engagement in the first eight weeks and a long tail thereafter.
The conservative rule that follows the research is to pay monthly for the first three months of any new subscription. The annual discount on those three months is small enough that the loss is bounded. If at month four you still use the service consistently — if your bank statement looks the same as month one's, if your usage logs would show a steady pattern — switch to annual. The savings curve is largely the same, and you've protected yourself against the most common failure mode: prepaying for eleven months of a service you stop using by month four.
The harder case is what to do about an annual plan you signed up for under the spell of the original prompt, and that you have now stopped using halfway through. The standard answer is to wait it out — the money is sunk, the service is still available, you may as well use it. The framing matters. Whether you "use it" or not, the same money was already spent the day you signed up. The remaining usage is a forward-looking question independent of the past payment. If the service produces less utility than other things you could spend equivalent attention on, the answer is to stop using it now, mark the unused months as a small lesson, and pay monthly next time.
What Laibson and Samuelson and Zeckhauser collectively showed, across decades of replications, is that the friction of the moment is the strongest predictor of long-run behavior. Annual billing reduces that friction, on average, in favor of the seller. Monthly billing reintroduces it, on average, in favor of the buyer. Neither billing model is right or wrong in the abstract. The right model is the one that matches how you'll actually use the service — which is best discovered by paying monthly for a quarter and watching what happens.