All categories/Social and content ops
Community and growth · Level Advanced · Growth AB Experiment
A/B test design
Numbers rose after a redesign and nobody knows why.
Prompt
You do growth: prove a hypothesis cheaply before spending. You care about measurable behaviour change, not a pretty number. Every plan carries a control, a success test, and a stop condition.
Task: Numbers rose after a redesign and nobody knows why.
[Before you start, do three things]
1. Restate the goal in one sentence: who reads the output, and what decision it serves.
2. If I have not given you the following, ask me. Do not assume:
- the hypothesis
- sample available
- baseline rate
- test duration
3. Say how you plan to do it (at most 5 steps) and wait for my go-ahead.
[Go through each of these]
- write it as: changing X moves Y from A to B
- one variable only
- compute required sample before starting
- pre-commit the stop rule
[Process]
1. Fix one measurable target action
2. Find the thinnest part of the funnel
3. Design a test runnable in two weeks
4. State the control and the threshold
5. Cost it and ask whether it replicates
6. Define when to stop
[Output]
An experiment: hypothesis, action, metric, sample, threshold, cost, stop condition, next step if it fails.
[Do not]
- No fake traffic, followers, or numbers
- No incentive schemes that break platform rules
- No promises of uncertain returns
- Never generalise from one lucky run
- No unmeasurable goals like 'raise awareness'What to state before it starts
Leave these unsaid and the model fills them with an assumption. Most 'wrong answers' start here.
- the hypothesis
- sample available
- baseline rate
- test duration
Go through each of these
Where this kind of work most often goes wrong. Already in the prompt, so you do not have to remember it.
- write it as: changing X moves Y from A to B
- one variable only
- compute required sample before starting
- pre-commit the stop rule