Track 11 · Experiments and A/B Testing · Lesson 4 of 14
Conversion rates and their uncertainty
Estimate a conversion rate with its standard error and a confidence interval (Wald and Wilson), and plan how many users are needed for a given precision.
A dashboard shows yesterday's conversion at 3.4% and today's at 2.6%, and someone asks what went wrong. With a few hundred buyers a day, swings like this are expected from chance alone. Knowing how much a rate wobbles is the foundation of every A/B test.
A conversion rate is a proportion: , the number of users who converted divided by the number of users. If each user converts independently with the same underlying probability , the count follows a binomial distribution, and varies from sample to sample around .
Included with a Formula Quest membership
It has 7 graded exercises with feedback, hints and worked solutions. After it you will be able to:
- Compute a conversion rate and its standard error
- Build Wald and Wilson intervals and choose between them
- Plan the number of users needed for a target margin of error
- Judge whether a daily fluctuation is just chance