You should only have two problems in your life as a website owner- driving visitors to your site and making them buy/enquire more. I’ve covered a lot of research in my previous posts on conversion rate optimisation about different elements that can be tweaked to improve your conversion rate. Applying these changes to your website as “best practice” is a little like driving blind – it’s only after you crash or arrive at your destination safely that you really know the end result (I dramatize for effect here, of course).
If you have a reasonably high volume website, A/B testing is the answer to driving blind. It uses actual visitor data to test out the effect of changes on your website. Then it gives a statistically significant answer as to whether the changes bring about an actual increase in conversion rate or not.
For some truly mind-blowing examples of how effective this can be, check out www.abtests.com.
A/B testing involves running two versions (let’s call them A and B) of a page with differing elements. A/B tests are run in simultaneously – that is, half of the visitors to a page will see version A and half will see version B. The visitors are generally split 1 for 1 or randomly to maintain consistency. Once the test is complete, statistics on visitor behaviour are compiled to answer the simple question: did more people convert on version A or B? The information unearthed by A/B tests will tell you which version of the page performs better – the control (A) or the variation (B).
Again, I highly recommend browsing around www.abtests.com. Here are just a few of their successful case studies:
- WikiJob experienced a 34% increase in the number of clicks through to the PayPal checkout from their product page – just by adding a testimonial to the Paypal checkout section.
- Performable had 112.5% more clicks on “Free Trial” button by using a well-known image of The Beatles instead of a generic.
- The Corkscrew Wine Merchants had 148% more people adding a product to the shopping cart just by adding a red sticker advertising the special, adding the discount before the product name on the page title and capitalising their free delivery offer.
- CityCliq had almost 90% more people clicking through to see pricing plans by changing their headline from “Businesses grow faster online!” to “Create a webpage for your business”.
Is My Site Right for A/B Testing?
A/B testing takes the guesswork out of optimising your website to be conversion friendly, but it’s not for everyone. Here are a few questions that will tell you if A/B testing is right for you.
- Do you get enough traffic to get statistically significant data? Generally, you need at least a 2000-3000 monthly visitors to your site to compile enough data.
- Does yur website have a base level of user-friendliness? Simple questionnaires of random people will uncover the answer to this (if the answer is no, start with the basics).
- Do you get a lot of return visitors? A/B tests should be run on new visitors only, to prevent freaking our your return visitors.
How Do I Start?
The first step with an A/B test, as with anything you do with your website, is design. Specifically, look at what elements you will be testing and how you will change them. I’ll dive deeper into this in my next blog post – so stay tuned!