What is A/B Testing?
A/B Testing is part of the CRO field (Conversion Rate Optimization) that gathers all kinds of marketing tools and methods used to optimize websites and web campaigns.
The cornerstone of this test, also called Split Testing, appears in the comparison of the efficiency of several versions of a same web page in order to determine which one converts the better. The first goal being the conversion of visitors into customers by choosing the best appropriate title, the best picture, what colour of a call-to-action button suits the better…
With the A/B Testing, we try to find what is the most efficient, through experimentation and interpretation of results. However, several points are to take into account so you don’t obtain biased or wrong results..
A/B Testing: watch points
On one hand, let’s talk about the sample size associated to the duration of the test. The tested versions should receive a representative traffic in order to have a fair comparison basis. In the same vein, we advise you to define a segmentation: what person do you take into account? Everyone or the one that have followed a particular path (that has done X previous actions, that has clicked on a particular page etc.) ? Last but not least, one of the major errors to avoid is testing several variables at once, which could wrong the test: testing only one will allow you to have a better understanding of what content strategy you should conduct.
If you are interested in A/B Testing, it is very likely that you already master these subjects. But have you ever paid attention to the loading time of your tested versions?
A/B Testing & Web performance : a bias risk
We already talked about it : only 1 more second in loading time can cost up to 7% the conversion rate. When you carry out an A/B Testing, according to the variants you have chosen, you may decline the loading time of one of them. Examples are numerous (change of a picture, use of a video etc.).
If any vigilance is up, the deterioration of the loading time can be significant. Then, there is little doubt that the corresponding version will have its conversion rate tumbling down and the interpretation of the results will naturally make us abandon this variant. The problem is that this abandon will come up for wrong reasons!
Let’s take the example of these 2 photos:
Intuitively, we’ll think that the right image will give us best results. Without making any change, still we’ll have the opposite results! There is a simple explanation to that: the right image is 100 times heavier than the first one and is going to penalize excessively the user experience.
By optimizing the image on the right, without touching to its quality, we can obtain a similar loading time on both pages. The results will be then completely different.
The impact of loading time on conversion is widely recognized. By doing an A/B test, we are generally out of the usual workflows (the picture to test isn’t from the usual graphic designer, we are not going through our CMS controls but directly through the A/B Testing solution, etc.) and we are then more exposed to web performance loss. You should absolutely keep this watch point in mind !
DareBoost & Unbounce have partnered up
In the approach of web performance integrated to A/B Testing, we have partnered up with Unbounce, landing page A/B Testing specialist. We are glad to see their interest in this issue.
By announcing you this partnership, we take an important step forward and hope to be able to give you integrations of DareBoost.com very soon to enhance at most your A/B Testing experience.
Meanwhile, you can test directly on www.dareboost.com the different versions of your A/B Testing!