This week I wanted to talk about something that I had heard a lot about, but just recently started implementing.
Before I get started, here’s a quick definition of split testing:
Split testing (also referred to as A/B testing) is a way of conducting controlled, randomized experiments with the goal of improving a website metric or advertisement, such as clicks, form completions, or purchases (Optimizely).
Since split testing can be used to improve conversions, I decided to start testing it out myself. I ran various tests on my Facebook ads as well as my landing page. I choose to test different images in one test, audiences in another, as well as short versus long copy. I also tested different versions of my landing page to see which would convert better. As I was only able to test one element at a time for best results, stay tuned for a future blog post on my findings!
Here are a few more ideas of the different elements you can test:
- Videos, images and visual elements
- Call to action text
- Layout: size of buttons, form elements, placement
- Audience demographics
- Placement: mobile vs desktop, right hand column
Before you get started with your testing, keep these tips in mind to ensure you’re getting the best results.
1. Pick your goal
What metric do you want to focus on with your testing? Are you wanting to increase conversions? Link clicks? Time spent on page? Be super clear on what you are wanting to improve before you begin your testing so you are able to make sure you are setting up your testing the most effective way.
2. Only test one element at a time
In order for split testing to be effective, you can really only test one element from the list above at a time. For example, you could decide to run an advertisement and test the image to see if a different image converts better. To find out which image actually performs better, you need to keep all the copy, targeting, headline, etc. the same.
3. Test both variations simultaneously
In order for split testing to work, you need to run both variations of your ad or webpage at the same time. This way you can be sure that the results you are getting aren’t based on timing.
4. Understanding statistical significance
Statistical significance can be a bit complex – essentially, it means that you want your results to have a confidence level of 95% or higher. The higher the percentage, the more sure you can be of your results and which variation of the test to choose. For more information on calculating statistical significance, check out HubSpot’s blog about it.
5. Test again
Once you have the results of your first split test, don’t stop there. Even if you think your landing page, advertisement, web pages, etc are doing well, split testing is a great way to see if they could be doing even BETTER. Ever heard the phrase “you don’t know what you don’t know?” Well, it applies here. Split testing gives you valuable data you can use to make informed decisions. You might even find out that what you thought was “good” actually could be great.
I hope this helped to shed some light for you on the mysterious world of split testing! If you’d like to chat more about it, feel free to book a free consultation call and we can discuss your digital marketing goals.