What should I wear this morning? Stitch Fix can tell you thanks to its algorithm
Stitch fix leverages its algorithm to revolutionize not only the way customers shops for clothes online, but also how they perceive fashion trends and identify their own style.
Stitch Fix is an fashion e-commerce startup founded in 2011 by Katrina Lake with the mission to help customers “save time, look great and evolve your personal style over time”. [1] By blending the human element of personal stylists and advanced proprietary algorithms, Stitch Fix is able to predict how your style will evolve over time and which items you would be happy to have in your wardrobe, providing each customers with a truly personalized experience.
The clients – currently both women and men – sign up on the website or app and fill in a survey aimed at discerning fashion preferences. The profile that each customer fills in collects more than 85 data points, to make the process as efficient as possible and the items selection as accurate as possible. A woman is asked if she is a mother or currently pregnant, as well as her due date. She also hands over her dress, waist and bra size; her age, job and location; parts of the body she would like to flaunt or downplay; and answers to more-abstract questions, such as whether she likes taking risks. [2]
After signing up and building their profiles, customers start to receive customized boxes of clothes composed for them. Each Stitch Fix order (called a Fix) is processed by 5-10 styling algorithms. The company relies heavily on its team of data scientists, counting nearly a hundred of them, and has algorithms for logistics, inventory management, inventory procurement, product design, demand estimates, etc. on top of its most famous algorithm to match a customer to a set of 5 items he or she may want to have. [3] Stitch Fix is also able to create new products combining the most popular design elements form existing customer orders, and to process the comments from clients to create items missing from the market.
However, in order to deliver a superior customer and item-customer matching experience, Stitch Fix recognizes the importance of human judgement when comes to identifying fashion trends and customer style preferences. The final result is an accurate mix of data output and interpretation by actual stylists. This is what probably represents the “secret sauce” of Stitch Fix and what made the young startup so successful in doing what other startup had failed to do with only analytics.
Similar start-ups, from rival Trunk Club to the cosmetics specialist Birchbox, have found a market mailing consumers items and offering free returns. But many such start-ups have had trouble keeping costs down and retaining customers after the first purchase. [5] According to the company’s founder, Katrina Lake, “success comes down to delivering what consumers want: making it easier to shop” [5] and also keeping innovation as the highest priority in the company.
Stich Fix counts now more than 2 million active clients, who drove nearly $1 billion in sales last year. Growing its sales more than 10 times than in 2014, the company has become one of the biggest online clothes retailers in the United States [3], with a $120 million initial public offering last November.
Colson – Stitch Fix Chief Algorithms Officer Eric Colson – says “We consider our algorithms to be composed of both expert human judgment and machine learning. When clients request a Fix, the selection is narrowed down and ranked using a set of algorithms but final selections are always made by a human.”
Looking ahead, the main challenges for Stich Fix are the following:
- Style standards are relative and differ based on where someone lives. The more customers will adopt the service in different locations in the world, the more the algorithms will have to be “smart” in adapting fashion elements to the specific context. For example, “edgy” may mean something different in Utah versus California. The on-boarding algorithm consequently may have to include even more than 85 data points, with the risk to create a barrier to customer acquisition.
- Traditional retailer could follow suit and erode market share. Indeed, while they don’t have in house the structure or value system to support the change required, traditional retailers are racing to acquire or partner with startup that can help them compete. Nonetheless, the tech-retailer giant Amazon is aggressively pushing expansion in online apparel and can pose a significant thread to Stitch Fix.
- Finally, the corporate harvest of data about our bodies, including our faces, voices and fingerprints, also is raising privacy concerns about how much sharing is too much in service of better-fitting clothes. “The very first data breach of someone’s jean size is going to be very treacherous for that company,” said Pam Dixon, founder of the World Privacy Forum, a consumer research group. “This extraordinary profusion of health-related data is not covered under anything. You have to rely on the privacy policies of the company, and it’s really not enough.” [4]
Sources
- Company website
- https://fashionista.com/2018/01/katrina-lake-stitch-fix
- https://www.forbes.com/forbes/welcome/?toURL=https://www.forbes.com/sites/veronikasonsev/2018/03/15/can-algorithms-replace-humans-at-stitch-fix/&refURL=https://newsroom.stitchfix.com/media-coverage/&referrer=https://newsroom.stitchfix.com/media-coverage/#71e195e84041
- https://www.washingtonpost.com/business/economy/companies-race-to-gather-a-newly-prized-currency-our-body-measurements/2018/01/16/5af28d98-f6e8-11e7-beb6-c8d48830c54d_story.html?noredirect=on&utm_term=.e709b3ca919f
- https://www.nytimes.com/2017/05/10/business/dealbook/as-department-stores-close-stitch-fix-expands-online.html?rref=collection%2Fbyline%2Fmichael-j.-de-la-merced&action=click&contentCollection=undefined®ion=stream&module=stream_unit&version=latest&contentPlacement=1&pgtype=collection&_r=0