The setup
- Goal: find out whether a question hook beats the current opening.
- Variants: the control and two question hooks.
- Metric: purchases, with hook rate as an early signal.
The numbers (made up)
- Results needed per variant: 30 purchases.
- Recent cost per purchase: about 25.
- Budget: 30 x 25 x 3 variants = 2,250 over 10 days.
Rules
- Stop a variant if it spends twice the cost per purchase with no purchase.
- Verdict: a variant wins if it has at least 20 percent lower cost per purchase with the agreed number of purchases.
If the money is short
- Test two variants, or test hook rate as the metric, and say it is an early signal.
The test budget sizing guide explains the method.










