We’ve been following Stitch Fix for some time now, and for good reason: the publicly-traded personal style service was named “Most Innovative Company” by Fast Company, and has continued to spawn a slew of competitors in both tech and retail.
The Fast Company cover featuring Stitch Fix CEO Katrina Lake.
When you talk to employees, they usually point to the company’s treasure trove of data as one of the reasons for its success. For the unfamiliar, Stitch Fix works like this: when you sign up, you’re asked a series of questions about your size and style preferences. After initial sign up, you’re able to share more about what you like through “Style Shuffle,” a fun, interactive feature that allows you to “thumbs up” or “thumbs down” everything from individual clothing items to entire outfits.
Screens from Stich Fix’s Style Shuffle in prototype.
Over time, the service learns your exact fashion preferences and sends you the right mix of clothing right to your door.
“Style Shuffle was a huge unlock for the company,” says principal product designer Torunn Skrogstad. The simple Tinder-like feature started as the simplest of prototypes and, after some iteration, is now the prime data collection method of one of the most successful startups of the decade.
It’s a testament to the power of the design process: how the right workflow of testing, listening, and iteration can have a profound business impact. We asked Skrogstad to share how the game-changing feature came to life.
1. Start simple
The idea for Style Shuffle began in a “crazy eights” brainstorm, an exercise commonly used in design sprints. In a “crazy eights” brainstorm, participants are encouraged to come up with eight new ideas in eight minutes, sketching each one out as an illustration.
(Check out this guide to running your own remote design sprint.)
It was from there that the idea of “Style Shuffle” first took hold.
The first version was released in Facebook Messenger. It wasn’t branded or very well-designed.
The beginnings of Style Shuffle on Facebook Messenger.
“It was a little slow, [and you’d have to] wait for the thumbs to appear. So, the design challenge was how to make it faster,” says Skrogstad.
When the team tested this concept with actual users, Skrogstad says they saw some promising signs. “The engagement was higher than what we expected. There was probably something there, so we pulled it onto our own platform”
2. Do qualitative user research
With the early feedback from the initial test, the team put it in front of more users.
(For the real-deal experience, walk users and stakeholders through your project using InVision Cloud prototypes.)
“I don’t want to ask ‘Imagine you like this red coat, what would you do?’ You’ll never get learnings from an artificial scenario,” says Skrogstad. So she mocked up a new, more branded version of the feature in Keynote, exported the animations as gifs, and moved the files into an InVision prototype.
Sample screens from Style Shuffle.
“I found it easy to export two different gifs for the red jacket: one for yes, one for no. We did that for 10 jackets and 10 pairs of jeans. We watched them rate and we got some interesting feedback,” said Skrogstad.
3. Keep it simple
“Style Shuffle” built a game out of the essence of Stitch Fix—one that gave the product team mountains of sophisticated data to work with, based on billions of interactions.
The goal was to keep the feature working smoothly at all times. “We heard from users that they wanted things like ‘super likes,’ ” says Skrogstad. “But we see such great engagement, we’re working hard to not screw that up.”
(Want to create a feature as successful as Style Shuffle? Get started with InVision Cloud.)
And more engagement means more perfectly personalized items. The first delivery of Stitch Fix clothing is a big brand moment for customers. If a box of clothing arrives that isn’t right for them, they likely aren’t going to come back for more. “We don’t want to send people in Florida a heavy jacket,” says Skrogstad.
4. The “unlock”
The collective data points go beyond the individual user. They create a “style map” for the entire customer base, which allows Stitch Fix to become more predictive with the information they glean from Style Shuffle.
“Now there’s a whole team using the data we collect for the rest of the service,” says Skrogstad. The brand is finding gaps in the clothing market and sourcing new clothing based on the data.
The gamification of Style Shuffle also increases the metabolism of the company. Rather than wait for people to receive and send back clothing, the company can learn more about their customers’ tastes quickly through clicks—instead of waiting to receive returned merchandise.
“We’re kind of reinventing the way people buy things. It’s reinventing retail, but it has a very human element to it. Everyone needs to wear clothes, and we spend a lot of time working on making it easy to find those things instead of having to go to the store.”
The fact that it all began as a sketch and an InVision prototype isn’t lost on Skrogstad.
“I was sitting in Copenhagen fiddling with these [thumb icons] and I had no idea this would be a huge unlock for the business.”
by Sean Blanda
Sean Blanda is the Editorial Director for InVision. Previously, he was the Editor-in-Chief of Adobe’s 99U and a founder of Technically Media. He currently resides in Philadelphia.