Treat big decisions as hypotheses

Most decisions are bets

Every significant decision in an early or growth stage company rests on assumptions. Assumptions that customers will pay this price, this channel will reach them, a study will produce certain results, or a message will be persuasive. Usually the assumptions are unspoken. The decision gets made on conviction, and when results disappoint, the team argues about what went wrong without having agreed on what success would have looked like.

The remedy is familiar. Lean startup practice made it standard advice for product decisions: build a small version (an “MVP”), measure what happens, learn, adjust. What is less common is applying the same discipline to the decisions that sit with the CEO: a regulatory path, a new channel, a senior hire, the design of a first study. Those are the decisions most often made on conviction, because they feel too big or too urgent to test.

They can be tested. Treat the decision the way a scientist treats a hypothesis. Name the assumption. Say what you expect to see if it is true. Set the result that would make you stop or change course, before you test it. Then test it.

What changes when you do

Three things change, and in our experience they matter more than the test itself.

First, the argument moves earlier. A team that has to agree on what success looks like before a decision has the hard conversation while it is still cheap. Disagreement surfaces as a difference in predictions, not as blame after the fact.

Second, the decision has an end. A hypothesis carries its own review date and its own exit rule. When results arrive, the question is not whether the decision was good, reopened from the beginning. The question is whether the team saw what it said it would see.

Third, stopping becomes normal. Most companies keep weak projects alive not because anyone believes in them, but because no one defined in advance what failure would look like. When researchers at Bocconi University taught this approach to founders in a series of randomized trials in Milan, Turin and London, the clearest effect was on stopping: founders who learned it were more likely to end weak projects, and to end them sooner.

Why it reduces decision-making burden

A CEO’s attention goes to every decision still open and every project still alive. Decisions with a review date stop getting re-litigated in every meeting. Projects with a defined failure point end when they reach it, instead of lingering. Both return attention to the decisions that need it.

A small, bounded test is also often how a one-way door becomes a two-way door: commit a little, learn, then commit fully. (See The few decisions you can’t undo.)

Why health innovation makes it harder

In consumer and digital health, decisions fall into two groups that take very different amounts of time to test.

Many commercial decisions can be tested in weeks with real customers: price, packaging, onboarding, channel, which benefit to lead with. Evidence decisions take a clinical study and often a year or more: whether the product improves a health outcome, whether it is safe, and what the company can claim.

Not every commercial decision is fast. When a sale depends on a prescription, a new reimbursement code or enterprise adoption, the commercial test can take as long as the study. Those decisions need the same planning as an evidence decision: the assumption, the threshold and the date, set well ahead.

Two mistakes follow from mixing them.

The first is waiting. The team holds commercial decisions until the study reads out, even though price, channel and onboarding do not depend on the study’s result. A year passes with little learned about how to sell the product.

The second is substituting. A concept test shows that people respond strongly to a promised health benefit, and the team treats that as support for the benefit. It shows that people want the benefit. Only the study can show that the product delivers it.

The fix is simple to state. For each assumption behind a decision, ask whether a commercial test or a clinical study can answer it. Test the commercial assumptions now, while the study runs. Keep commercial results out of decisions about what to claim until the study supports the claim.

A simple discipline

Before the next significant decision, write down four things.

1. The assumption. What has to be true for this to work?
2. The prediction. What will you see if it is true?
3. The threshold. What result would make you stop or change course? Set it before you test.
4. The date. When will you look?

Then decide, and leave it alone until the date arrives.