Hypothesis
Turn product ideas into falsifiable MVP hypotheses.
Key takeaways
- The hypothesis is the unit of learning; vague hypotheses make fast execution produce only ambiguous output.
- Use the template to name customer segment, urgent problem, solution slice, observable behavior, time window, and reason.
- Classify the test as a problem, value, channel, or delivery hypothesis, since each needs different evidence.
- Rank signal on the evidence ladder: compliments are weak, while committed money, contracts, or repeated usage are strong to very strong.
The hypothesis is the unit of learning. If the hypothesis is vague, fast execution only creates more ambiguous output.
Hypothesis Template
We believe
[customer segment]has[urgent problem]. If we offer[solution slice], then[observable behavior]will happen within[time window], because[reason].
Four Hypothesis Types
| Type | Question | Example evidence |
|---|---|---|
| Problem | Does the pain exist? | Interview pattern, search demand, repeated workaround |
| Value | Does the solution matter? | Demo conversion, waitlist, LOI, paid pilot |
| Channel | Can we reach them? | CTR, reply rate, qualified traffic |
| Delivery | Can we deliver reliably? | Activation, support load, fulfillment time |
Evidence Ladder
| Signal | Strength |
|---|---|
| Compliment | Weak |
| Survey interest | Weak to medium |
| Demo booked | Medium |
| Manual workflow requested | Medium |
| Money, contract, or time committed | Strong |
| Repeated usage or renewal intent | Very strong |
Checklist
- The target customer is narrow enough to find.
- The problem is observable, not just assumed.
- The success behavior is measurable.
- The time window is short enough for a sprint.
- The team agrees what evidence would disprove the idea.