How long should an A/B test typically run to achieve a 95% confidence level?

• A. One week
• B. Two weeks
• C. At least four weeks
• D. Six months

Answer: (C)

Reaching 95% confidence in an A/B test relies on having enough data to overcome natural noise and to capture regular weekly patterns in user behavior. Short runs often don’t provide enough observations, so results can look convincing by chance or miss how behavior changes over weekends and different days of the week. Running for at least four weeks gives you data across multiple weekly cycles, making the performance estimates more stable and reliable. That makes the result more trustworthy without delaying decisions longer than needed. One week or two weeks usually don’t accumulate enough data for most sites to confidently claim 95% significance, while six months is typically longer than necessary unless you’re dealing with very slow traffic.