How to create better A/B tests based on User Research

A/B testing is great. But if you just go around testing every idea that pops into your head, you could be in for a lot of testing. And ultimately, bitter disappointment.

Karl shows AB-tests that really made a difference thanks to user research – defined as analytics, targeted surveys, user session recording, and user testing. He shows cases from car brands, web shops, luxury products, health products.

Drawing from real life examples he showed why putting big pictures in the header area is not always a great idea; or how to encourage people to enter personal details into forms by explaining how their data is going to be used; or why repeating call to actions at the end of a page makes a lot of sense; or why sometimes it’s better to push product categories instead of individual products on e-commerce websites. The talk is packed with lots of practical advice.

Nowadays when A/B tests are quite easy to do thanks to services such as VWO and Optimizely there’s not much stopping us form using this technique. However, it’s important to plan the tests beforehand. Every site is different, so there’s no point in copying tests from someone else.

Also, it may be helpful to gain some understanding of user expectations by making online surveys – instead of asking for opinions Karl Gills simply suggests asking the users: “What is the purpose of your visit today?”.