A/B Testing Definition

A/B testing, also known as split A/B testing is a method of comparing two versions of a webpage or application, variant "a", the control or current experience, is compared against variant "b", the variation, to determine which one performs better. AB testing is an evidence based practice, using data & statistics to validate new design changes and improve conversion rates. The objective of A/B testing software in web design (especially user experience design), is to identify changes to web pages that increase or maximize an outcome of interest (e.g., ad click-through rate)

A/B Testing - Step by Step

  1. Take a webpage or app screen and modify it to create a second version of the same page. This should be limited to a single change such as a single headline or button.

  2. Expose half of your traffic to the original version of the page (known as the control)

  3. Expose the traffic remainder to the modified version of the page (the variation).

  4. As visitors are served either the control or variation, their engagement with each experience is measured and collected in an analytics dashboard and analyzed through a statistical engine.

  5. Analyze this information to determine whether changing the experience had a positive, negative, or no effect on visitor behavior.

A/B testing allows developers to make evidence based changes to their user experiences as a consequence of collecting data on the results. This methodical approach dispenses with opinion in favor of empirical results affording developers enhanced insight into why and how certain elements of an application impact user behavior. Consequently, AB testing can be used to consistently and continually improve the user experience, thereby improving the conversion rate over time.