In the e-commerce world, when people discuss about customer acquisition, the first thought that comes to their minds is related to traffic or online marketing. These are very well known practices that bring more customers to a particular site by increasing the traffic through different strategies. These methods go from affiliations to e-mail marketing, search engine marketing, display marketing, re-marketing, white labeling and any other strategy, but the interesting point is that there is very little literature about A/B testing or UX (user experience). It seems that the power of A/B testing is undermined but the potential of its ROI is exceptional.
Even though this is a less common or discussed subject, the process of A/B testing can have a significant impact in your site and your consumer acquisition conversion rates. The concept is very simple and makes the customer choose the option he/she likes the most without telling him/her that they are actually helping you. The main point here is that most marketers spend a lot of time trying to grow in traffic -which implies a great effort in terms of budget- but do little A/B testing, which could increase sales 100% or 150% with little effort and money. The level of improvement will depend on how good or optimized is your funnel.
There are software on internet that make the process of optimization very simple, easy and seamless. The concept is not difficult:
1) You create two to five variations of one landing page or one form from your site that you wish to test.
2) Each landing page has to have only 1 element that will vary from version to version, for example the color of the call-to-action button or the picture in the background. It is fundamental that there is only one variation between each version because if there is more than one it will be impossible to find out which element was the trigger to the improvement.
3) The experiment will normally have to run for a couple of days in order to achieve a level of significance. Most of the software out there will tell you how many conversions or impressions are needed in order to have a relevant result. In each experiment, you can measure different results from click-through-rate to conversions or engagement.
4) Once you have run the test for a few days, there will be a winner and it is going to be the option that had a better response by the users. The software for A/B testing will normally show one version to each customer automatically until a sample of customers have seen the same number of different versions, and the results will became clear towards one of the options.
5) Once you have one or two potential winners (the closest second), you can start the same test with those top two pages until you find the one that is the most successful.
This process is almost infinite because it can be created for each of your landing pages, result pages, or any other part of your funnel. In the case of our company -which is a comparison site- our funnel has a Landing page, form pages, result pages and purchase pages. In each page, there are different call-to-action buttons to test or different possible layouts to test. Therefore, in this particular case there are probably over 10 tests of the funnel per area, and then there are re-testing of winners. Thus, we could be talking about over 100 tests or more.
In order to get deeper into the subject and the potential results, I am sharing below a current A/B testing that we are performing in our business for a particular landing page in our site, where we are testing three variations of the picture in the background.
As it can be appreciated from the Performance Summary above with only 331 unique visitors this test is showing that the engagement in the landing page improved 9.8% with the Bond Girl picture, 2.5% with Friends and 24.5% with Family. The original picture is just a car. The most interesting part is that if we decide to go with the Family picture, we will also gain a 46.5% CTR (click through rate) to the next section of the conversion funnel. Such result in itself is astonishing, as the effect of the family picture versus the original car one will give the site a equivalent boost of +46.5% more traffic at the original design and this just by changing one image. The relevance of the test it is at 90% now, therefore we need more time to have a clear picture of who the winner is, but once we have collected enough data, we will change the picture and go on onto a new test.
The best part of A/B testing is that customers tell you if they like green more than orange in a call-to-action button or if they prefer a picture of a blond girl or a family as the background without knowing. Many people may ask themselves why big sites like Amazon or eBay sometimes look ugly and the real answer is: This is what the mass market likes because those big sites are all about A/B testing.
