A/B Testing Strategies
A/B Testing is use to test campaign elements when your company doesn’t control the audience. For example, if I do an emailing to my customers, I control the audience so I don’t need to do A/B split testing. I use Head to Head testing because I can control who gets which email message. For websites, testing messaging, testing landing pages and other applications, we don’t control the audience. As a result, we need to use A/B Tests.
A/B Testing is simply a testing methodology. You can test as many items as you want. For example, you have a web page which isn’t engaging very many people. If it a source of friction because of its lack of performance. You call in 2 marketing specialists and they develop two alternative pages for you to evaluate. How do you determine if one is better than your web page, you do A/B split testing.
What you do is test each page, in your website, for a set period of time. Perhaps you measure your original page for 24 hours, then do alternative page 1 for 24 hours then the same for alternative page 2. You repeat this test day after day until you have exposed all 3 equally to your high value market. The length of the test is determined by the number of people visiting your site. I will give you information on how to determine sample size in another training session.
To develop an A/B test strategy, here is what you do:
- Determine what you want to test and, if needed, what the control [current] is.
- Develop a tracking system to measure each test element
- Test each alternative until you see which is the “winner”
- Use the winner
In my Northwestern digital marketing classes, we use A/B split tests to test 4 alternative messages for a blog the students write. To help them, I have developed a spreadsheet which you can download and a series of videos to show you free tracking systems and how to set up and evaluate an A/B testing system. This is what I use at Northwestern but you can use it for any A/B test you would like to perform.
Let me know if you have any questions or comments.