Setting up Tests
now that you understand what an A B test is. Let's talk about how to set one up. A B tests should follow the scientific method for many of you, this is probably a throwback to high school. So let's brush up on some of these concepts first, you should determine what conversion you want to improve. In my case I want to improve the conversion rate for my homepage. Second, develop a hypothesis based on the user testing you conducted earlier or your own experience or even a hunch. You should try to figure out why you think your pages not converting at the level you want it to be converting. What could be different about this page to make it work better for you. In my case I'm deciding the copy on my C T A button is a bit too aggressive. I think I can get more people to click with a softer message. The third step is to define more specifically what the control and test versions are. The control is typically the variation you have on your site already. In my case it's shop now the test is the...
version I think will improve conversion rates in this case I think changing shop now to find your design will improve conversion so I designate those as my control and my test. Finally, the last two steps are implementation and measurement in a second, I'll show you an affordable tool you can use to set up and measure A B tests on your site before I do that. A bit of inspiration. There's so many things you can test on your website. Here's a fun list of 10 things you can a B test. You probably will recognize some of these from previous chapters. You can test the C. T. A copy size, color placement, social proof. You can add or remove a trust seal. You can test a form that is one steps or two steps. You can test the visuals on the page, illustration versus photography, you can add or remove videos, directional cues, navigation, you can test the length of the form and finally you can also test the headline. Notice that really you can test nearly anything whether it's a design element or a psychological component like social proof. You can even test pricing here are some examples from real companies A. B. Testing different elements on their pages. Here's A C. T. A. Or call to Action test from one bounced in their control version. They're asking users to share their email address. In exchange for this book. In the test version you'll notice everything looks exactly the same as in the control version with one key difference. The call to action. The test is asking users to tweet in exchange for the book result Asgrade users for an email had a 24% higher conversion rate than asking them to tweet interesting Right now, let's take a look at a copy A. B test in this example This design agency, a. B tested the headline copy from a more generic message to something more specific that calls out, they help clients with online and offline marketing needs in this case, the test version one and led to an 8% increase in click through rate to other pages on their site. Finally, I'll draw from an example of my own. When I was director of marketing at Zillow Group, we may be tested one of our most important landing pages. You can see the two designs here. Note that for this test I decided to test the entire page and not just one standalone element. If you have a page on your site that you feel needs a complete overhaul, I suggest going for this approach rather than a B testing things. One by one result We lifted conversion rate by 200%. A similar thing happened to revenue. It was such an impactful test and a testament to the importance of A B testing and conversion rate optimization. Now that we looked at some examples, I'm sure your brain is full of ideas. Let's take some time to write them down on our worksheet. You'll notice that I added three A B test ideas for my own business. You can pick your favorite one and expand upon it a little bit by defining a hypothesis, a key metric or what the success mean for you. In my case it's improvement of conversion rate from my home page and finally a test and a control version. Next I will show you how to set this all up using a tool called optimizing. You might be already thinking yes A B. Testing makes sense for my business. It's something that I want to try with my website but how do I do it? So introducing optimized lee, this is one of my favorite tools that is affordable and easy to use to conduct a B tests on your site with very little help from the development team. So if you scroll down on the optimized the homepage, one of the best ways to know um get a sense of how this tool works is to choose right here. Marketing and growth in the in the role section and to select to try out the visual editor. So take one second to load. But basically this visual editor gives you a really really good idea of how optimized we would work for your business if you were to sign up for an account. So um this is the home page of what looks like an e commerce website. So you would basically select any element on this home page that you want to a B test as you can see. You can select almost anything, the background picture, you can select the branding up here, you can select the entire navigation, you can select different elements of the navigation, the text and then you can also select the call to action button here. So let's just start with the call to action button that is one of the more sort of elementary and essential A. B. Test that marketers um start dipping their toes with in terms of a. B. Testing for landing pages and for and for their first website. So if you click on this you'll see uh that it'll just ask you to basically change this element and what what you want to change it to. So let's say we wanna change shop now to find my style save. And so now uh you have the original which is this variation this shop now and you have variation number one which is fine my style. Um And from here you can very easily launch this test uh and make sure and then you would be basically a B. Testing this these two versions what we would call the original or the control or a version And then what we call the test or the B version which is variation number one optimized. The also allows you to add more than one variation. This is what's called multi variant testing and you're not able to really see it here in this in this in this editor view. But the other great thing about optimized lee is that you can also get a robust set of data back for every experiment. So you can really understand the performance of each variable when each variable when when the tests are statistically significant. Uh And then when it's time to start thinking of a new test to continue to optimize the performance of your site or of your landing pages. So uh as someone who has used this tool before, I definitely recommend it. Another tool that I won't do a walkthrough for but that I have used in the past and also recommend is google experiments um if you use google analytics, um it works seamlessly with google analytics, so it's something to consider. But I'm a bit partial to optimize lee and um highly recommended as a way to easily set up a. B. Tests for your landing pages or your website.