What is A/B testing?

• A. A randomized survey to gather opinions.
• B. A long-term observational study without a control group.
• C. A controlled experiment comparing two variants to determine which performs better on predefined metrics.
• D. A usability test with a single interface variation.

Answer: (B)

A/B testing is a controlled experiment that randomly assigns users to two variants and compares predefined metrics to see which performs better. The key is random assignment and a measurable outcome, such as conversion rate, click-through, or revenue, so you can determine if any observed difference is real and not due to chance. This setup gives a clear, data-driven way to decide between the two options, rather than relying on opinion or observation alone. For context, other descriptions don’t fit: a randomized survey concentrates on opinions rather than actual behavior on the product; a long-term observational study without a control group can’t attribute effects to a specific variant; and a usability test with a single interface variation doesn’t provide a direct comparison between two options.