Ad testing is the comparison of variants of copy, headlines and calls to action, with the aim of finding the one that produces a better result. The most common mistake is premature evaluation – with a small number of clicks, variants take the lead at random, and whichever moment is declared the end of the test gives a different answer. Therefore, always test only one significant change, let the test run long enough to reach a meaningful sample, and evaluate by conversions or their cost, not by click-through rate, which can rise even with a worse result. When variants are generated automatically by the system, supply enough distinct building blocks, otherwise the combinations will differ only in word order. Record the results so that the insights are not lost.
See also: Statistical significance of a test, Creative fatigue, Ad extensions.
