The cornerstone of an healthy organization is experimenting. It’s something we as humans have done since the beginning of time. Even as a child we experimented, our parents just called it ‘learn to ride a bike’. After some experimenting we learned to keep our balance, discovered new area’s on our bike and have been experimenting the rest of our lives.
This experimental thought is essential for the organizations we work in today. Without those experiments we will not be able to bike around and discover new truths, new worlds, new possibilities. We will just sit in a stroller and be pushed around by others. Luckily we do experiment and learn from our mistakes and successes. Organizations have done this for years. Vikings wouldn’t have discovered North America without experimenting. This experimenting has some risks, but the risk of not experimenting is even higher. Some organizations like to think they can avoid this risk by making decisions based on gut feeling. But the question is if they take the right decision, today we have tools and methods to support the experimentation and decisionmaking. Experimenting shouldn’t end up into a stomach ache…
When running a business it is important to know what the customer wants. A/B testing can help the business in this search. A/B testing means in short: randomly display 1 of 2 options to customers and see which option has the highest response. These 2 options can be for example a website banner with a different color, another font, placement of buttons, different art-work or button sizes. Everything that is related to the UI of a website or app. It is however not limited to a website. Some organizations have different types of commercials, banners or letters that are sent to customers. Even work-methods can be A/B tested.
But how to perform an A/B test? It all starts with an idea, a hypothesis. For example, imagine we have a blue button on our website and we think that a green button might have a higher click-rate. Therefore we randomly present a page with a green button and a blue button to our visitors. This random process is an important part of this process, we shouldn’t have any influence on this process. All the other circumstances should be identical. We can’t, for example, have the blue button on monday and the green button on tuesday because this might disturb the measurements. After a week of experimenting where we randomly had 50% of the users visiting a site with a blue button and 50% with a green button we can analyze the results. Important for this analysis is that we track which user is displayed which option and if the user has clicked or not. Now we can calculate the clickrate and make a decision. The green button works best with a high margin.
Now we can make the green button the default and start with the next experiment. How about moving the button from the left to the right….