What is A/B testing?

A/B testing, aka split testing, is an experimental method that compares the performance and behavior of two alterations of a product part or feature. They can be both new alterations, but mostly we are talking about the current version against a new one. Let’s think of the Google search bar and buttons:

As we can see above in the left version of the google search bar, its format is curved and on the left side, we have a rectangular form. So, before Google made those changes, they had to find out why they should do it, besides looking nicer and fresh for the product manager and product designer. This should affect positively their business, in this case, the search experience.

Why should product managers run A/B testing?

Building great products takes excellent work. When you want to create products that users will love to use, testing is one of the most critical things as a product person. 

Taking the previous example, let’s say you’re the product manager working on the google search bar, you don’t want to change it, just because you feel like it. Why should you do that? This search bar has brought Google to enormous success! 

So think of your goals as a product person for this essential feature. Let’s say, you want the users to find the right search result faster. In this case, we assume you already know the, for this to happen, you will need the users to provide more words in the Google search.

So you decide to work with the designers on finding out how the search bar UI will influence it. A/B testing would be a good validation tool for this. You can show the new variant of the search bar to a specific percentage of the users from the city of Lisbon. Then compare the results that you get with the results from the other slice of people in Lisbon that were using the current search. Now you can grab that data, and make a valid decision, that you know will bring value to your users from Lisbon, be it to change it or not (you’re also better prepared to answer challenging questions). 

How to start A/B Testing?