How To AB Test
Stop losing clicks and customers. Here is one example of how the conversion rate on a page was increased from 1.4%
with Visual Website Optimizer
. Put your science hat on. Try Out the Free Plan
Every proper experiment starts with a little hypothesis. In the example below, I wanted to drive more traffic at the end of an article towards a promoted product. With concept A being the initial control variation, I hypothesized that concept B would perform better due to the following applied GoodUI ideas:
- 09Telling Who It’s For instead of targeting everyone. (“Hey Designers” reference)
- 10Being Direct instead of indecisive. (Removed the “Want to tell ...” reference)
- 11More Contrast instead of similarity. (More depth, more color, more tone)
The Setup Screencast
Here is a 10 minute video on how to actually setup your first A/B test with Visual Website Optimizer.
Some Awesome Results
After running the test for 5 days with just over 1400 visitors, both Visual Website Optimizer and myself became quite convinced that the second variation was successful. In fact, concept B was +3.2% higher in absolute terms than concept A (or +222% in relative terms). The software actually determines the winner for you based on some statistical analysis, so you don't have to do relevancy calculations. With this, I was ready to stop the test and implement the new variation across the full site, potentially generating three times the referrals to the product - a pretty decent result.
You Can Do The Same For Just $49 / A Month
To get started with A/B testing, I’m recommending the $49 a month Small Business Plan from Visual Website Optimizer (yes, it’s an affiliate link). With that plan you can run A/B or multivariate tests on 10,000 visitors each month, receive support from their awesome team, and track results in real time. I find that this plan is also perfect for running a couple of solid relevant tests each month. Sure they also have a free plan, but that is only limited to 1,000 visitors which can run out pretty quickly and some tests actually require more data to be relevant (especially if you will be running more than two variations). The cost is a very good investment and can be quickly offset once you realize how much more you could be getting out of your existing visitors. Good luck ...