Want to become a better designer? Start by rethinking your A/B tests
As a designer, you might come up with a change to a product and think to yourself “Any rational human being will like this change.” But then, through an A/B test you find out that… your users don’t like the change. Design intuition might chalk this up to humans just being weird, while data might suggest that there was something inherently wrong with your design.
For a long time, Mike Davidson, InVision’s VP of partnerships and community, thought this way: viewing data and intuition as opposing forces in product design. But then, back in 2012, he read something that changed the game for his career: “Building Websites with Science” by Peter Seibel.
In the post, which ran on Etsy’s blog Code as Craft, Seibel takes a diplomatic approach on how to use data to build and redesign websites. It taught Davidson that he can use data as a tool to help inform design rather than control it.
For example, that change you assumed your users would love might not be performing as expected not because it was a bad design or the wrong decision. It could have been because they had a hard time reading the accompanying text or even that they simply didn’t have an open mind when trying out the new feature.
From the article, Davidson learned to look at data as an indicator rather than a decision maker.
“[The article is] a nice reminder that creativity is still a very important part of the product development process and you can’t always boil everything down to just A/B tests,” Davidson says.
To put it into perspective, think about climbing a mountain: Data can tell you the most efficient path to the top, but if all you do is follow the results of A/B testing all the time, you may miss that there is a much more magnificent mountain in the distance that you’d be better off on in the long run.
Powerful prototypes in minutes
Show distinct flows for A/B testing with prototype sets in InVision Cloud.
Davidson says the article helps get designers, who may have no interest in data, excited about data-driven experimentation.
“It has made me look at each decision and ask, ‘Is this something we should test or not?’” Davidson says.
Davidson said the article also helped him better collaborate with his engineering colleagues because he could better understand why they’d be resistant to installing changes without A/B tests. He learned that to get engineering buy-in, he should frame design directives as not just a subjective good idea, but one that the team has tested and proven best for the product.
In sum, Davidson says he learned from Seibel’s article that data is there to help you do a better job, but that it’s not there to keep you from doing the best job that you could do—a reminder we all need in this increasingly metrics-driven world.