How To A/B Test A Page Within The Episerver Editor
As of Episerver 10.1, it's now possible for you to start A/B testing your Episerver page. A/B testing, also known as split testing, is a technique of showing two different variations of your web pages to your users at the same time. Split testing works by randomly assigning and displaying a variant to a user, measuring if the visitor clicks on a button etc.. and then after a period of time comparing the statistics between the two variants to figure out which version converts better. Following this process can help to dramatically improve your sales funnel. The good news.. setting this up in Episerver is really easy!
How To A/B Test Your Episerver Page
First, find a page that you want to perform a test on. Make your changes as normal and when you are finished, in the normal publishing drop-down in the top right-hand side of the main navigation panel, select'A/B Test Changes'. This will load the 'A/B' testing screen. The first thing you will be presented with is a comparison view of the original page and the new variant. To start a test you can click the 'Start Test' button at the top right-hand side of the main content panel. It will fail though unless you set a conversation goal so you will need to scroll down to the test options section. Before we get into the configuration details, it's probably a good time to talk about statistical significance. Statistical significance is the figure you need to reach to prove that the difference in conversion rates between a variant and the baseline is not due to random chance. You should keep this fact in mind when you run tests because to reach good statistical significance, your tests need to be viewed by enough people. If you run tests for a few hours, for example, you will never reach statistical significance. When you run tests you really need to have them visited pretty frequently, usually for a month minimum in order to be confident the results are accurate. This is where the Episerver options come in handy. Now you understand about statistical significance, the options seem a bit more obvious. They help you target your tests to reach this goal. Conversion Goal: This is the action the visitor must perform in order for the test to be proven. Adding new goals will require a developer to write some code, so if you want a specific conversation goal, it's worth talking to them to understand what it possible. At the time of writing, the 'Landing page goal' is the only included goal that comes with Episerver, but that will change in the future. When this goal is set, Episerver will class the test a success if the site visitor visits the chosen page. Participation percentage: This is the total percent of people who will be given the test who visit that page. Setting it at 100% will mean, a 50/50 test. 100% is the best option for reaching statistical significance the fastest, however, you may not want everyone to participate in your test. If you're doing the test on the homepage, for example, you might only want to run your test against, 10% of the people who visit your website in case the test has a negative impact on the conversation. These are the sorts of trade-offs you need to consider when you start A/B testing. The more people you send to your test, the quicker you will know the outcome, however, is your test dramatically reduces conversation you could loose a lot in sales, so it's up to you to find the right balance. From my personal experience I've usually gone for 100% but the decision is all yours. Test Duration: How long the test will run for. Based on my previous experience I would say 30 days is a minimum unless you have an extremely high traffic website. The more people you can participate in your test every day, the more you can trust the result.
A/B Testing Takeaway
A/B testing is an important tool to help improve your sale funnel and improve conversion. As I'm hoping you can see, A/B testing in Episerver is super simple. You make your changes as normal, but, instead of publishing the results. You publish them as an A/B test instead.