Hypotheses
What it means
Growth experiments turn guesses into evidence by testing one clear change at a time.
The useful version of this topic is not theoretical. It should change something visible: the customer you choose, the words on the page, the offer in the email, the onboarding step you remove, or the metric you watch every week.
A good test is whether a founder could use this page and do something before lunch. If the answer is no, the topic is still too abstract.
Actions
Start with the smallest action that creates contact with reality. For early distribution, reality usually means a buyer, a user, a search result, a conversation, or a payment.
- Write the hypothesis before launching the test.
- Define the success metric.
- Record what changed and what you learned.
Measure
Measure the signal closest to customer progress. Vanity numbers are allowed only if they predict something better later.
- Experiment velocity
- Win rate
- Measured lift
Template
Hypothesis: [if we do X, Y should improve] Audience: [segment or channel] Primary metric: [one number] Guardrail metric: [number that must not get worse] Decision rule: [ship, iterate, or stop]