Risk transfer

Route-to-market gets argued with adjectives — "safe", "leaves money on the table", "market risk" — when every structure on the table is just a payoff transform on the same uncertain year. Draw them on one axis and the argument changes: the floor is a put you paid for in upside share, the toll is certainty priced at the whole distribution. The quotable numbers fall out — what upside you sold on average, what protection you bought, what the fees were once separated from the risk terms. Binding probabilities get an honesty check: two numbers don't pin down a left tail, so when the lognormal and normal fits disagree materially, the verdict says "12–22% depending on tail shape" instead of pretending.

Named simplifications, so the tool can't oversell itself: the toll is modelled as a fully fixed payment — availability incentives and performance obligations are out of scope. Insure is a clean indemnity with an attachment point and a limit — no deductibles, no basis risk. Every structure is evaluated on the same merchant year, so counterparty credit risk — a floor is only as good as whoever wrote it — is explicitly not priced. And it's one representative year: multi-year sequencing, debt covenants and path-dependence belong to a cashflow model, not this comparison.

The tool prices the trade; it never says "choose the toll" — that depends on a risk appetite no simulation should claim to know. Sources: the real-options framing of revenue floors as puts; Douglas Hubbard on calibrated 90% intervals (How to Measure Anything).