The Stitch Fix story: changing the way millions of people dress with data

“Style is very personal. It has nothing to do with fashion. Fashion is over quickly. Style is forever.” — Ralph Lauren

Personal style is not only how we dress but also how we present ourselves to the world. It reflects a lot about our personality. In the traditional retail model, customers go into a store, rifle through the racks of available clothes, and seek out the items that best fit their overall style. Whether or not the store has these items is a bit of an unknown, but one the customer has to accept.

Stitch Fix wanted to change that.

By gathering data on the core elements of what a customer considers stylish and distilling it down to an algorithmic view, Stitch Fix is able to deliver personalized recommendations for each of their customers. This fundamentally changes the buying experience, and it’s proven very attractive to customers. Stitch Fix has grown from a startup to a multi-billion-dollar public company in just seven years.

With data-driven personalization, AI machine-learning systems, and an understanding of customer psychology, Stitch Fix is upending the traditional retail clothing experience.

Your personal stylist fits in a box

Founded in 2011 around the idea that technology and data science can help make personal styling easier and more accessible, Stitch Fix has raised $42.5 million in funding over four rounds. More than half of that funding came later in the game, via a $25 million C round in 2014.

Stitch Fix, via Twitter

Stitch Fix was founded by Katrina Lake and her sister as a personal-shopping website, through which a $20 styling fee bought customers hand-picked garments, delivered right to their homes. Even during the early stage, Lake understood the need to track customer preferences in order to be successful, though her data collection and parsing efforts were far from sophisticated. In those early days, Lake and her sister collected information via a simple SurveyMonkey form.

From those humble beginnings, Stitch Fix has grown revenue to a recently reported $1.2 billion, largely because that rudimentary data operation has become much more complex. By focusing on gathering and analyzing extensive customer data and working to automate their processes whenever possible, Stitch Fix has grown from a modest, women’s only subscription-box service to a retail powerhouse that now also serves men and kids, and has special options for plus-size, petite, and maternity clothes.

How StitchFix finds the perfect outfit for everyone

From the beginning, Lake and her team at Stitch Fix understood that to be successful, they needed to choose garments—on the first try—that were as close to perfect as possible. Getting it right quickly makes the customer happy, builds on Stitch Fix’s reputation as a style authority, and encourages repeat business.

They found that the best way to do this was by collecting and analyzing data on their customers’ style and preferences.

The first step in this process happens as soon as the customer creates their Stitch Fix account. Customers who set up an account, either through their email or by connecting to Facebook, are led through an extensive questionnaire called the Style Quiz.

Example quiz options, via Stitch Fix

The information collected at this stage is incredibly important because it underpins the entire relationship between the customer and the brand moving forward.

Once Stitch Fix has that information, they feed it through their recommendations engine, a series of styling algorithms that help pick items based on customer preference, and then forward it along to a human stylist. That stylist makes the final decisions about which clothing to send to a customer based on the items presented by the algorithm and puts together a box of five pieces. Using human decision-making augmented by machine algorithmic choices is part of what makes the Stitch Fix model more successful. (More on this in the next section.)

Human input means Stitch Fix can be an arbiter of fashion, rather than just an amplifier.

Customer-specific personalization is how Stitch Fix differentiates themselves from many of their competitors. Instead of sending everybody a box with the same set of clothing the company thinks will be popular, Stitch Fix personalizes recommendations based on the information they have on their customers. Everyone’s box is different and personal to their own specific style and taste. And it’s working: Stitch Fix is currently one of the most popular online subscription-box retailers in the United States.

Most popular subscription boxes by monthly visits, via Statista

Stitch Fix is third only to Ipsy, the cosmetics subscription service, and to meal-kit service Blue Apron. With over 3 million monthly visits, they’re also leaps and bounds ahead of Bespoke Post, their nearest semi-competitor among the top ten.

By collecting qualitative feedback on each box they send and updating customer preferences accordingly, Stitch Fix ensures that boxes get better over time. That increases their retention rates and the potential for future monetization. Customers who receive items they love in each box are more likely to stick around longer and keep more garments.

How Stitch Fix melds data and taste

With such a focus on data and algorithms, it’s important to note that Stitch Fix also employs human stylists to make the final choices for their customers. The combination of data science and human stylists helps to ensure that Stitch Fix makes the right decisions for their customers. It also shows how they use data science to automate processes for their team.

Every step of the customer journey is mapped, analyzed, and optimized by algorithms.  Stitch Fix employs more than 80 data scientists who work on everything from clothing design to package fulfillment and logistics.

Everything they do is built on the idea that data-driven decisions help not only their customer but also the entire Stitch Fix team.

And Stitch Fix is just one of many fashion companies leveraging this kind of technology to automate their internal and external processes. Technology is changing how clothing is designed and manufactured.

Technology’s impact on fashion, via CB Insights

When a stylist receives the recommendations from their AI counterpart, the majority of the heavy lifting has already been done. Stitch Fix’s human stylists just need to make sure the clothes are the best fit for the growing customer profile they have. It’s an objective layer on the totally algorithmic decisions made by the machine-learning magic Stitch Fix is running all the time. This humans-in-the-loop approach turns AI into intelligence augmentation rather than allowing machines to completely replace humans. The human input means Stitch Fix can be an arbiter of fashion, rather than just an amplifier because it allows people the chance to inject sometimes surprising or counter intuitive style suggestions that that data might not reveal on its own.

The CB Insights graphic above shows just a handful of ways technology is being used in the fashion industry to help automate different aspects of the business. Stitch Fix also uses deep learning and algorithms in their packing and shipping. From pick paths to warehouse product availability, everything is tracked and analyzed. This helps Stitch Fix’s team stay productive in the face of an ever-changing retail landscape.

The psychology behind Stitch Fix’s 65 percent first-month retention

Setting up their Stitch Fix accounts takes time, and the longer a customer has used the service, the better the matches have likely become. Each time a customer receives their Stitch Fix subscription box, they’ll be charged a $20 styling fee, and that’s just the beginning. Once the customer has their five items to choose from, the real decisions need to be made. They can either keep the items in the box or pack them back up and return them to Stitch Fix.

This is a great example of how customer pain is used in the subscription-box model. Stitch Fix customers will need to confront either the immediate pain of boxing up unwanted items and mailing them back or the delayed pain of the purchase showing up on their credit card bill at a later date. Their aversion to either type of pain will inform how the customer moves forward. Either way, Stitch Fix has already received their $20 styling fee. Even if the customer decides to send back every item in the box they still have revenue coming in.

This is also important from a retention perspective. While Stitch Fix does have higher-than-average retention rates in the first six months, any technique they can use to keep their customers around longer is important.

Monthly customer retention, via Goodwater Capital

If the customer decides to keep any item from their subscription box the $20 styling fee is credited to their purchase. That acts as an additional incentive to keep at least one of the items that were provided to them in the shipment. Customers can always update their account to a different schedule but the styling fee remains the same.

Personalization is a powerful tool in the eCommerce market, but you need to have enough information to make the right decisions for your customers.

This focus on the psychology of their customers, combined with Stitch Fix’s ability to make their subscription more and more personalized with each new box, acts as an incentive for the customer to keep more items with each new shipment. The initial questionnaire and subsequent feedback provided for each subscription box, coupled with the perceived pain of sending back unwanted items, makes it more enticing for the customer to continue their relationship with Stitch Fix.

There’s also a data lock-in element at play. Because Stitch Fix is constantly building and adding to the style profile it keeps on each customer, and recommendations get better over time as a result, there is increasing pain for customers to switch to a competing service.  It’s kind of like switching doctors—that gets harder the longer you’ve seen the same physician because they know your history and idiosyncrasies and you’re comfortable with their care. Do you really want to go through that whole process of forming a relationship again with a new person? The same concept is at play with Stitch Fix. Moving to a different stylist means starting over from zero with someone who doesn’t know anything about your style preferences, so the longer you stay, the more painful it becomes move.

Key takeaways

Personalization is a powerful tool in the eCommerce market—even more so for subscription-box services like Stitch Fix. With a rigorous focus on using data to make the right decisions for their customers, Stitch Fix has been able to grow from a modestly funded startup to a retail company with staying power.

► Collect all the data

Stitch Fix knows a lot about their customers, and I mean a lot. The company has combed through questionnaires and feedback to make sure that they have plenty of data on hand. You need to have enough information to make the right decisions for your customers.

► Analyze everything

Data is worthless without analysis. You should be consistently evaluating and reevaluating your customer information to ensure that it’s up to date and correct. Learning about your customer’s preferences is the best way to provide them with a product that will make them happy.

► Don’t forget the human element

Machine learning, AI, and algorithms are amazing, but without a human being to interpret the information provided, you’ll miss out on a lot of nuance. Technology can’t have a feeling about what’s right, or make a decision that’s a little bit out there or adventurous; it can only provide the best possible match based on what information it has.

► Let customers decide

Stitch Fix makes it easy for their customers to give feedback on pretty much everything. Giving your customers the power to decide what works best for them will help your business provide exactly what makes them happy.

► Automate intelligently

Giving your team the tools they need to be successful will help your company grow. When you leverage technology the right way, it gives your people the freedom to make decisions based on each customer interaction. They can spend more time helping the customer, and less time working with your product. Think of automation as augmentation, not a way to replace your human teammates.

[optin-monster-shortcode id=”nwltspncrcafgjpcetis”]


<em>Hmm, looks like this article is quite old! Its content may be outdated, so comments are now closed.</em>