How are users typically divided in an A/B test?

• A. A random 50% split between the two variants
• B. A 70/30 split favoring the control
• C. Division by geographic region
• D. Division by device type

Answer: (A)

In A/B testing, users are assigned to each variant at random, typically with an equal chance to see either version. This random assignment creates two groups that are comparable in characteristics, so any observed difference in outcomes can be attributed to the change being tested rather than who happened to see it. An even 50/50 split maximizes statistical power for a given total sample size, making it easier to detect real effects in metrics like conversion rate or engagement. In practice, you might use different splits for specific reasons (such as reducing risk with a new feature or gradually ramping up exposure), or you might stratify the randomization to balance certain factors across groups, but the standard approach is a random 50/50 division.