You’ve probably heard before that correlation does not imply causality (or causation, if you prefer), but do you really understand what this means and why this is important in conversion optimization?. We’ll try to clarify this for you.
Correlation vs Causality
Let’s take a look at the following graphic (sorry, no idea about the source).
As you can see there is a very strong visual correlation between the IE market share and the murder rate. The less the market share the better (better = less people murdered). So, could you deduce from the graphic that banning Internet Explorer would stop all murders in the country? Obviously, the reason is no (you could argue that the despair provoked by using IE could make some people crack but I don’t think this accounts for all the murders in the country 🙂 ). IE market share is not the cause of the decrease in the murder rate.
So, does this graphic bring any useful information at all? Well, no. The fact that the two variables (IE share and murder rate) are correlated could indicate that one causes the other but also that they are both caused by a third factor or, as in this case, not even that. It may be just a coincidence (i.e. both variables depend on different external factors that just happen to take place in the same period). At most, since causality always implies correlation, you could take a correlation relationship between two variables as a good sign that it’s worth to investigate a little bit more the origin of this correlation in the hope of finding a causality relationship between them.
Switch to WordPress safely while keeping your design and content
Improve your SEO today and boost your site speed by converting your pages into HTML, CSS, and WordPress standards. With zero technical knowledge required, you’ll only pay for what you need.
Correlation vs Causality vs Conversion Optimization
Ok, I understand the difference but why does this matter when optimizing my WordPress site?
Because, even when people know the difference at the abstract level, are too often fooled to optimize the site based on correlation relationships. Imagine that you decide to change the theme of your WordPress site this month and your sales go up a 10%. You may think that the theme change was a good thing when maybe if you had stuck to the original theme your sales would have gone up a 30% instead. There are a myriad of reasons that may explain this 10% increase beyond the theme change (maybe you have a seasonal product, maybe an important blogger wrote a review of your site,…). Correlation may kill your site instead of helping it
So, what should I do to improve my conversions?
Aim at identifying causality relationships. For that you need to conduct controlled experiments using a tool like our WordPress AB Testing service. Do you want to see if changing a theme will increase your sales? That’s how you do it: randomly show the site with the new theme to half of your visitors and the site with the old theme to the other half and see which half buys more from you.
This way you eliminate external factors or better said, the external factors still exist but since they are the same for both groups of visitors, the result is not affected by them. For instance, the group seeing the new theme can show a sales increase of a 20% while in the other group the increse is of only 10%. The fact that both groups (even the one where you didn’t modify anything wrt previous months) buy more is due to external factors but the fact that the group seeing the new theme brings you more money is only due to the fact that new theme is indeed better (i.e. it has a better conversion rate) so you can now be sure that switching to this theme is the right thing to do!