A/B testing, also called split testing, is a method for comparing two variants of something to learn which drives better results. You show version A (the control) to one group of visitors and version B (the variant) to another, then measure which produces more conversions, clicks, or whatever goal you set.
How it works
Traffic is split randomly so the two groups are comparable. You change one element at a time, for example a headline, button color, or form length, and let the test run until you collect enough data to trust the result. Testing a single variable keeps the outcome interpretable, since a multi-change test cannot tell you which change caused the lift.
Statistical significance
A result is only worth acting on when it reaches statistical significance, commonly a 95% confidence level. That means there is roughly a 5% chance the difference is random noise. Ending a test early because an early lead looks promising is a frequent error, since small samples swing wildly. Most tools recommend a minimum sample size and a run of at least one to two full weeks to cover weekly patterns.
Why it matters
A/B testing is the backbone of conversion rate optimization. It replaces guesswork and opinion with evidence, so changes to a page are grounded in how real users behave rather than what the team assumes they prefer.