Skip to content
writing

Every Program Thinks It's the Exception. Most Aren't.

The reasons a project feels exceptional are rarely the reasons that matter to its odds.

EvaluationGovernanceHypothesisLast tended · 2026-07-30

Confidence: hypothesis. The failures below are drawn from published research on decision-making and clinical development, not firsthand deployment. The discipline at the end is an argument, not a shipped system. Argue with it.

Ask any team why their project will beat the odds, and they will have a good answer. A better design. A smarter approach. People who have done this before. The answer is usually sincere and usually detailed. That is the problem. The harder you look at what makes your project special, the less you notice what it shares with every project that failed the same way.

The clinical why

Drug development runs on a single number. It is called probability of success, often written PoS or PTRS. It is the odds a company puts on a program: the chance this drug reads out positive, reaches approval, and launches. That number decides which programs get funded and which get killed. It moves real money.

The base rate behind that number is public, and it is harsh. An industry study by BIO and Informa, covering 2011 to 2020, found that a drug entering Phase 1 has about a 7.9% chance of ever reaching approval. Most programs that start do not finish.

Now sit in a few portfolio reviews and watch what happens. Nearly every program is rated above its own base rate. That is not possible for a whole portfolio at once. If most of your programs are above average, the average is wrong. It is the same as a town where everyone says they are an above-average driver.


This pattern is one piece of a longer treatment. The full essay is issue 4 of Stage × AI, a series walking the entire clinical-trial lifecycle stage by stage — what each stage really does, where AI helps, where it must not go, and one buildable pattern per stage:

full essay
Probability of Success: The Number That Moves the Money

Evidence in, evidence out. Corrections welcome.