Obama and Split Testing

A few days ago I came across a very interesting article on the Optimizely blog, a tool that facilitates A / B Testing, about how the Obama team managed to increase the conversion rate of a landing page for donations in 2007 thanks to a simple experiment.

This simple experiment that consisted of testing different variations of images and calls to action on the landing page allowed them to change the course of the campaign.

The landing page sections to be tested were called "Media" (for the top image) and "Button" (for the button with the call to action).

During the experiment, 6 different types of "Media" (3 images and 3 videos) and 4 buttons were tested, giving a total of 24 combinations. The experiment was carried out using the Google Website Optimizer.

The buttons were all the same format, just varying the call to action, the following calls to action were used: "Sign Up", "Sign Up Now", "Learn More" and "Join Us Now".

You can see the different images and videos that were tested during the experiment by clicking on the following link: Splash Page Experiment

Results

The metric to measure the success of the campaign was the percentage of visitors who subscribed, the winning combination was this:

The conversion rate of the winning version was 11.6%. what represents an improvement of 40.6% Regarding the conversion rate of the original landing page of nothing more than 8,22%

It is estimated that thanks to the increase in the conversion rate, an extra 60 million dollars in donations were raised.

It should be noted that Obama staff members had a preference for videos over images and considered using them on the landing page, this would have been a huge mistake since during the experiment the videos presented the worst performance.

The morale

Many times decisions are made based on assumptions and in the worst cases are based solely on the personal taste of the highest ranking person.

Thanks to the great arsenal of tools we have today, many even free, it is at least silly to make decisions in this way.

With a little more effort you can make decisions based on hard facts rather than your own preferences.

And what experience with Testing can you share? 😉

You can see the original article here: Optimizely Blog