Stitch Fix recently filed to go public, giving people the first look at the company’s health, unique characteristics and competitive advantages. While many consider Stitch Fix yet another subscription box company, it is much more complicated—and fascinating—than that. By looking at how Stitch Fix contextualizes itself, uses data and humans in tandem, and is expanding its distribution and verticalization, it becomes increasingly clear that Stitch Fix is a new type of company that belies all previous labels.

The setup

The way Stitch Fix has contextualized itself is built on the premise that many of the advances digital commerce companies have made are not actually the most important advances. The company argues that even though offline retail is shrinking overall, ecommerce—where sales are still small overall but rapidly growing—still lacks an important personal touch. This creates an opening for Stitch Fix.

From its S-1:

The first wave of eCommerce companies prioritized low price and fast delivery. This transaction-focused model is well-suited for commoditized products and when consumers already know what they want. However, we believe eCommerce companies often fall short when consumers do not know what they want and price and delivery speed are not the primary decision drivers. There is an overwhelming selection of apparel, shoes and accessories available to consumers online, and searches and filters are poor tools when it comes to finding items that fit one’s style, figure and occasion. eCommerce companies also lack the critical personal touchpoints necessary to help consumers find what they love, further depersonalizing the shopping experience.

This is a direct shot at Amazon, even though the company is never mentioned. Stitch Fix believes that there is a massive amount of commerce that is not driven by price or speed, which are the two attributes that Amazon has always optimized for. Stitch Fix says that none of that matters if you can’t get the right products in front of the right people.

Data, humans and personalization

While humans and data are often discussed as rivals, Stitch Fix was built on the premise that the two supercharge each other, while delivering significant improvements to the shopping experience. Stitch Fix likely has the best data science team in fashion and apparel, which is currently composed of less than 200 individuals. It also has a styling team of over 3,000 people that work with the data and turn it into outfits for the company’s 2.2 million customers. Importantly, given women’s is Stitch Fix’s biggest business, over 86% of its employees and 55% of its management team is female.

Customers can talk to stylists like human beings, rather than checking boxes, and then Stitch Fix can come back with recommended products. The result of this human and data relationship is that Stitch Fix is able to push the envelope of personalization because of its “data network effect.” This means that the more people that use the service, the faster its data improves, which creates a better shopping experience and so forth. Stitch Fix collects 85 data points from each client and gets direct customer feedback from its shipments 85% of the time.

Stitch Fix, in this vein, is very similar to how Netflix uses existing demand signaling (users watching shows) to create new supply (new shows that it has a hunch people will like). Yet since Stitch Fix customers voluntarily contribute their data to help improve the service, its data is accurate to the degree that Facebook’s is because its users do most of the work. Therefore, Stitch Fix is well positioned to capitalize on the fact that humans will remain a big part of the shopping experience for the foreseeable future.

Distribution

One of the least talked about lessons of modern retail is that most distribution is not inherently good or bad. Ecommerce, wholesale, shop-in-shop, retail, subscription boxes and many other forms don’t have any specific properties that make them succeed or not. Only when applied to a specific business do some make more sense than others. A big lesson from Stitch Fix is that subscription boxes have not failed, but that other attempts struggled to put the right stuff in the box. Customers mostly care what is in the box more than they care about the box itself. The same could be said for retail stores or ecommerce: if the right stuff isn’t there no one cares about the channel. One can see Stitch Fix’s success in its 86% repeat purchase rate in 2017. People clearly like the products it suggests.

If Stitch Fix can continue putting the right stuff in front of the right customers, it is well placed to take advantage of many different types of distribution. While Stitch Fix didn’t specifically mention any offline retail ambitions in its filing, it seems like a logical progression that it could experiment with in the future.

The company also believes that the power of this data expands well beyond just selecting products for shoppers. It uses data and algorithms to “predict purchase behavior, forecast demand, optimize inventory and enable us to design new apparel.” For Stitch Fix, data is not a feature—it’s culture, and the results speak for themselves.

Private labels and manufacturing

Private labels are one of the fastest growing aspects of Stitch Fix’s business. Like many retailers have in the past and modern companies like Amazon and Dia&Co are doing now, multi-brand retailers are able to leverage their sales data to see where there is demand but little supply, and spin up labels to fill these holes. Over 20% of Stitch Fix’s revenue comes from private labels, which it calls Exclusive Brands. This number will likely grow over the coming years. It probably won’t hit 100%—that would diminish the selection that customers love and its stylists need—but it wouldn’t be surprising if this number hit 50% sometime soon.

One of the little covered parts of the company’s filing was the disclosure that it recently bought “knitting, cutting and sewing assets in Pennsylvania to experiment with making very small quantities of apparel to test with our clients.” It says it has “no plans to manufacture apparel in any meaningful quantities and anticipate that we will continue to rely almost exclusively on third-party vendors to supply our merchandise” but this could change—quickly. I would definitely keep an eye on its increasing verticalization on this front, which would give the company more more control in both design and production, in addition to capturing more margin as a result.

So what is Stitch Fix?

All of this raises one question: what type of company is Stitch Fix? Part of it functions like a retailer, with a 700-brand assortment, thriving private labels and a range of product categories. Part of it functions like a tech platform, with its data network effect, its constant user feedback, and its deep technological roots. Yet part of it is increasingly looking like a vertical product brand, with a strong consumer-facing brand, many of its own products, and a small but expanding supply chain that it controls.

Stitch Fix seems like an early but promising example of what I’ll call a Vertical Platform. It has the internal dynamics and attributes of a tech company like Netflix or Facebook; the multi-brand assortment of Amazon and other third-party retailers; and the singular brand and verticalization of a company such as Zara. While none of these alone are entirely unique or differentiated, together they create a package that is both desirable and defensible. So far, Stitch Fix has executed this synthesis better than anyone else.