For example. You may make an ucat guess about what the outcome would be of changing the time you send welcome emails. Similar to setting a goal. Your hypothesis should be s.M.A.R.T. (specific. Measurable. Achievable. Relevant. And timebound). In this case. Your hypothesis could be “sending welcome emails within 10 minutes of a user joining will increase email open rates by 6% over the next three months with the new user segment.”Split each segment into an “a” and “b” test groupNow that you’ve form your hypothesis. Split the subscriber segment in two: an “a” group for your control group and a “b” group for your test group.
Split the segment equally at random
Split the segment equally at random to ensure the results aren’t skew one way or the other. The easiest way to achieve random group selection is to use an email service provider (esp) that has built-in a/b testing.Assess if each group is large enough to provide statistically significant results to ensure the most accurate data. If the groups are too email list small or not vari enough. The test will be prone to just reflect the results of randomness. Whereas a larger group will increase the accuracy of results by rucing the probability of randomness.
A good starting size is usually
A statistically significant group is determin by a few factors and a lot of math. If you’re not a statistician or just don’t like doing math (because who does?). You can easily find the right size by using an a/b test calculator. A good starting size is usually at least 1.000 subscribers. But again. That can be lower or higher depending C phone number on the test and the subscriber list.Create “a” and “b” test assetsTo test a specific aspect of your email. Create two variations of the same email with just that single element chang to reflect your hypothesis.Rip curl. Australia’s leading surf brand. Uses dynamic content to drive conversions.Case study