Stop 09 · Growth & ExperimentationTopics · 3/10

Writing a Testable Hypothesis

In short

"Let's try a new banner" is a wish. "If we show the delivery time, then orders rise because uncertainty drops" is a hypothesis. Only the second one teaches you something.

A properly formed hypothesis states the change, the expected effect, and the reasoning — 'If we do X, then Y will happen, because Z' — turning a vague idea into something you can actually disprove.

The purple blob kneels between two tomato beds, one with bigger tomatoes, writing in a notebook.

Think of it like this

The tomato bed experiment

A gardener says: "If I add compost to one bed, its tomatoes will be bigger, because the soil is poor." One bed gets compost, the other doesn't. A month later, she weighs both. The "because" tells her what she learned either way.

Watch it happen

Building a hypothesis piece by piece

Observation

Change

Effect

Because

Measure

1/5Observation: 38% of users leave the cart page. Interviews say they're unsure about delivery charges.

The big ideas

What you need to know

If, then, because

State the change, the expected effect and the reason. The reason is what you learn.

Name a number

"Increase by 3 points" can be proven wrong. "Improve engagement" can't.

Base it on evidence

Good hypotheses come from data, research or user quotes, not random hunches.

Be ready to be wrong

A clear hypothesis can fail clearly. That's useful information too.

See it in code

Walk through the code

text

1Hypothesis: Delivery fee upfront2Because: users say surprise fees made them quit3If: we show the delivery fee on product pages4Then: cart completion rises by at least 3 points5Metric: cart completion rate (primary)6Guardrail: average order value must not drop >2%

1/3Name it, and give the evidence-based reason first.

A simple template you can paste into any experiment doc. Every line should be filled in before the test starts.

Save yourself the pain

Common mistakes

Oops

Writing "Test: new checkout design" as the hypothesis.

Do this instead

Add the expected effect, its size and the reason behind it.

Oops

Choosing the success metric after seeing the results.

Do this instead

Write the metric and threshold before the test starts.

Your mission

Write three hypotheses

~20 min

Quick check

Which is a testable hypothesis?

Go deeper

1Rewrite 3 vague feature ideas ('let's add gamification') into testable hypotheses.
2Practice stating the 'because' — your underlying belief about user behavior.
3Learn to pre-register what result would make you abandon the idea.