Daily Insights

How Stitch Fix Uses Data to Power Personalized Loyalty

By: Zoey Tang | Jul 08, 2019

This past month, personal shopping service Stitch Fix announced that it beat earnings and sales expectations in the fiscal third quarter of 2019, with 3.1 million active clients in the third quarter. In an environment where many retailers face growth challenges, Stitch Fix has thrived—in part, thanks to its sophisticated predictive algorithms.

Over time, consumers have shown increasing loyalty to brands who magically get them and put content or goods in front of them that they love. The strikingly accurate recommendations provided by brands like Spotify and Netflix have demonstrated the incredible power that can be unlocked with the right data science team. However, Stitch Fix is the first brand in retail to unleash this potential: the CEO of Stitch Fix, Katrina Lake, employs one of the biggest data science teams in the industry to create the predictive algorithms that keep users happy. 

Through a combination of the user’s Style Profile, merchandise information (e.g. color, length, price, brand), previous interactions with the service, and a plethora of other data points, Stitch Fix can confidently recommend a black leather motorcycle jacket to a customer who typically likes conservatively cut blazers. The method seems to be working: over 80% of users return for a second “fix” within three months. 

Stitch Fix’s data-first approach is innovative in the retail industry. In fact, only a quarter of the brand’s specialty retail counterparts actually deploy the data they collect to create a personalized experience, as shown in Gartner L2’s Data & Targeting Report

As the cost to acquire users goes up, it’s become increasingly important for brands to focus on customer loyalty. Retail brands can learn from Stitch Fix’s statistics-driven program by recognizing the powerful impact that effective personalization has on shopper satisfaction. Most importantly, brands should recognize that personalization requires steps beyond data capture—even the biggest troves of data lack power unless they are organized and analyzed in the right way.