The Tuition Concept: Pricing Your Failed Bets Instead of Hiding Them
A failed experiment is not waste if you booked what it taught you. Tuition is the practice of naming the cost of a losing bet, recording the lesson it bought, and never paying for the same lesson twice.
Published August 21, 2026
Every operator has a drawer of experiments that did not work, and the instinct is to shut the drawer. Tuition is the opposite instinct: name what the failed bet cost, record what it taught, and file it where you cannot pretend it never happened. A loss you booked as a lesson is not waste. A loss you deleted is waste you will repeat.
A failed bet is information you paid for
The search campaign we killed spent about six hundred dollars and returned no customers at a price we would pay. Framed as a mistake, that is money down a hole. Framed as tuition, it bought a durable, specific answer about where a brand's acquisition does not work — an answer that now steers the next dozen decisions. The money is the same either way. The bookkeeping is the difference between an expense and an asset.
The discipline is in the prediction
Tuition only works if the bet was defined before it ran. Because every action here is born with a metric, an expected result, and a due date, a miss produces a sharp lesson rather than a fuzzy feeling. This is what keeps “it was tuition” from becoming an excuse: you cannot retroactively decide what a loss taught you if the hypothesis was on the record before the money moved. Contrast the sunk-cost trap, where undefined bets get thrown good money after bad because no one wrote down when to stop.
Never pay for the same lesson twice
The payoff compounds only if the lesson is un-loseable. On an append-only ledger, tuition lines persist: the next time a similar bet is proposed, the record is right there saying we already paid to learn this. That is the mechanism that turns a scattered history of experiments into an institution's memory. A team without it re-enrolls in the same course every year and calls the bill a surprise.
Why we price failure in public
Publishing tuition — anonymized, rounded, never tied to a real brand's private numbers — is a trust move. It says the scoring is real, because you can see it land on losses. It is the honest counterweight to the eleventh question: we do not claim our agents only win; we claim we know, to the dollar and the date, what our losing bets cost and what they taught. That is a more valuable thing to be able to say.
Questions founders ask
- What is "tuition" in an operating ledger?
- Tuition is the cost of a bet that did not pay off, recorded as the price of a specific lesson rather than erased as a mistake. When an experiment misses the number it promised, the money it spent bought information: this angle does not convert, this channel does not scale here, this offer does not hold. Booking that as tuition means the loss produced an asset — a lesson you can point to — instead of just a hole in the budget.
- How is this different from "fail fast"?
- "Fail fast" is about speed; tuition is about bookkeeping. Failing fast still lets you forget what the failure taught, repeat it next quarter, and pay again. Tuition insists the lesson is written down against its cost, in a record you cannot quietly delete, so the same failed bet cannot be re-run as if it were new. Speed without a ledger just means paying for the same lesson faster.
- Doesn't calling losses "tuition" just excuse bad decisions?
- Only if the lesson is vague. Tuition is disciplined precisely because the bet was defined before it ran — a metric, an expected result, a due date — so a miss produces a specific, falsifiable lesson, not a comforting shrug. A loss with no pre-stated hypothesis teaches nothing and should not be dignified as tuition. The rigor is in the prediction that came first.
- Why publish your tuition instead of hiding it?
- Because a track record with no visible losses is not credible, and because the lessons are genuinely useful. A ledger that shows what we paid to learn — anonymized, directional — is stronger evidence of a working method than a wall of wins. Hiding losses protects a demo; publishing them, honestly and without exposing client data, is what makes the wins believable.