A/B testing is a method of comparing two variants of a chosen element of your communication or campaign (the creative, headline, CTA, landing page layout, audience or targeting strategy) by running them side by side to comparable groups of users. The winning variant is chosen on a hard result against the goal: CTR, conversion, CPA or ROAS, not on opinion. It is one of the basic optimisation tools and usually an ongoing process, in which successive tests gradually improve the messaging, the creative and the user experience. It works in both performance and branding campaigns, and especially well on landing pages, where even a single better headline or CTA can noticeably lift conversion. It is worth distinguishing it from multivariate testing, which examines several variables at once; A/B isolates a single element, so you know exactly what decided the result. (Compare: Split Testing)