Six great tips for split testing
22 February 2011
Split testing shouldn’t be a mystery to any marketer looking to make a difference. Simply put, split tests allow digital marketers to fine tune and improve conversion rates. They do that by testing two versions of a thing (page, subject line, form, whatever) against each other and measuring the results.
A subject line is a great example – getting it right can mean the difference between poor and fantastic results.
Another example might be the humble “add to basket button”. If there’s a reason that Amazon are the kings of ecommerce, it’s because they split test. Their (not so little) yellow button has been split tested to the n’th degree, like everything else you see on amazon.com.
So how do you do it?
Here are six tips to get you started:
1. Conversion goals. Know them. If you’re not absolutely crystal clear on what constitutes a conversion, you’re wasting your time split testing. If you want people to get in touch, it will be your contact forms. If you want people to download your report, it will be the download link or thank you page. Get to know these goals and the paths your website visitors or email subscribers use to get there.
2. Write your split test list. You’re looking for anything that, with an improvement, will result in more conversions. Buttons and links, forms, layouts, copy – if I have to interact with it in order to get to your conversion goal, it’s fair game.
3. Think like a user. Consider each of the elements on your test list and how they might affect the user journey. Are any of them blocking or frustrating your users? Are they obvious enough? Or too distracting? Are you asking too much of the user? Or failing to excite them?
4. Hypothesise before testing. Eg. “I wonder if some users aren’t finding this button - If this button were bigger, it would be more obvious and would enable more users to click through to conversion. When you’re clear about the hypothesis, test only that one factor – in this instance, button size. If you test any more than one variable at once, you won’t be able to trust the results. If you want to test button colour, or position, that’s another test.
5. Measure and understand. The end result of any split test is a set of results. If you did it well, it should be as simple as two numbers. Clicks for version A against Clicks for version B. If the difference is significant, you know what to do.
6. Rinse and repeat . Split testing is a habit, it doesn’t stop.
The great thing with split testing is that you can do as little or as much as you want. If it’s a subject line for an email campaign, that’s easy to test, with big rewards. However you approach it, remember that every tiny increase in conversion ratio is persistent and cumulative. Doubling your conversion ratio is effectively the same as doubling your traffic.