How to make your website a conversion machine with A/B testing
by Katelyn Fogarty
Test, test, test, everyone has an opinion of what will work best when it comes to websites but you can’t argue with data. We know marketers love data! (cute marketing pick up line, “I love you, like a marketer loves data!” )
Testing is a huge part of my everyday process on acquia.com. I currently have 8 active tests running for various parts of the website. I wanted to take you through my process and tools to help make your site a conversion machine. I am running Visual Website Optimizer to do a/b testing, multivariate testing and split URL testing. There are other services out there that do similar things with varying levels of implementation and price such as Optimizely, Google Website Optimizer and Vertser. Drupal also has many modules for testing. You can pick and try any of these as well.
What to test?
The list can go on and should also change to meet your needs. If your goal is to increase lead form fill out you should be thinking of placement of your form, length of your form and copy before and after your form. You need to make sure when you are doing your testing you are comparing your changes to a control simultaneously. Meaning if you test CTA text one week and different CTA text the next week your results could be skewed by standard fluctuations in business. To get the best results have your test run against a control at the same time. (Most A/B tools will do this by default.)
This is a good list to start thinking about:
- Headline copy
- CTA copy
- Button CTA color
- Form on the page
- Position of form on the page
- Site menu items
- Picture of a person on the page
- More or less content on a page
- Adding and removing form fields to increase conversion
- Fonts and font colors
- Background colors
Once you’ve decided what you are going to test and how you will be setting up and tracking your test, create one!
Tips from Kate:
- Make sure you let your test run until you have significant data, don’t make your conclusion too early! Here is a tool to calculate “statistical confidence” http://visualwebsiteoptimizer.com/ab-split-significance-calculator/
- Once you’ve found a winner doesn’t mean you can’t test it again with different variations, keep coming up with ideas and test them!
- Make your tests consistent site wide, if you are only testing elements on one page you will be all set but if you are testing menu item changes make sure that change happens everywhere to not confuse or frustrate your users.
- Show your tests to new visitors, this will ensure you don’t interrupt the standard flow for your returning visitors.
- Don’t let your gut or others opinions out weigh test results. Results don’t always show the most pleasing combinations but could better your conversion rate. Don’t reject results but do another test with parts of the previous test, maybe by trying combinations you’ll find a pleasing yet well converting option.