A decade ago, Stitch Fix, like so many DTC companies, expressed confidence in data's ability to boost sales through a level of personalization that traditional brick-and-mortar stores and maybe even many online retailers couldn't accomplish.
In her first conference call with analysts in 2017 after the apparel box retailer went public, founder and then-chief executive Katrina Lake explained how her idea — using a combination of algorithms and human stylists to help customers find clothes — is more convenient than shopping at physical stores and solves the "enormously difficult problem that e-commerce is not well suited to address."
The company has made great strides in 10 years. The U.S. apparel subscription e-commerce segment represents a sliver of the overall pie, about 1% of the total U.S. apparel and footwear market, but Stitch Fix's share of that subscription slice has remained steady at 58% since 2019, according to Coresight Research. In 2020, the company's overall share in apparel and footwear was about 0.5%, up from 0.4% the previous year, Coresight also found. In its last full fiscal year (reported about a year ago, in a period dominated by the pandemic), Stitch Fix's adjusted net revenue rose 11% from the previous year to $1.7 billion, though net loss reached $67 million; in its most recent quarter, the company regained its footing somewhat.
"Stitch Fix has a very good business model, but they've had some bumps in the road, certainly," Shawn Grain Carter, professor of fashion business management at the Fashion Institute of Technology, said by phone.
A disruptive idea?
Stitch Fix's concept seems simple.
When clients sign up, they fill out a questionnaire regarding their size, fabric and style preferences, and budget. The answers, plus subsequent information regarding purchases, returns and other feedback, are used by the company's tech and human stylists to help decide which five items will go into a box or "fix." Customers choose how often or seldom they receive a fix, with shipping and returns at no charge. If they choose nothing, they surrender a $20 "styling fee" and must return the box within three days; if they keep anything that fee goes toward their purchase price.
"This is a consumer-centric, client-first model that leverages technology and data science to help millions of clients discover and buy what they love in a totally new way," Lake said on that first call, according to a transcript from Seeking Alpha, adding later, "Our team of over 3,400 stylists combined algorithmic recommendations with their judgment in human connection to deliver a one-to-one personalized experience that is scalable, efficient and effective."
In order to deliver on that, though, Stitch Fix's business has become more elaborate.
To widen its share over the years, the company has added men's and kids apparel to its repertoire and expanded to the U.K. To glean more data from customers, it's added interactive features like "Fix Preview," which allows customers to select or reject items before their fix goes out, and "Style Shuffle," where customers swipe yes or no on fashion that does or doesn't appeal to them. (Other styling box services, including Nordstrom's Trunk Club, have similar features.)
However, a few other recent changes there suggest that, nearly four years from its IPO, the company's model may be struggling with scale, efficiency and effectiveness as it collides with some realities.
In April, for example, Lake took many observers by surprise with her announcement that she would step down as CEO in August to make way for Bain consultant and Stitch Fix President Elizabeth Spaulding to take the wheel. Spaulding, who arrived at the company last year and was Bain's global digital chief, has no experience in apparel or as a chief executive.
To FIT's Carter, that suggests an emphasis on efficiency at the expense of optimal operations. "Retail is in it to make gross margin profit, but you have to have a fashion merchant running the business," she said. "I don't care whether it's Stitch Fix or brick and mortar or a catalog or a rental company. You have to have a merchant making these decisions, not a management consultant, not a financial consultant."
"I don't care whether it's Stitch Fix or brick and mortar or a catalog or a rental company. You have to have a merchant making these decisions, not a management consultant, not a financial consultant."
Shawn Grain Carter
Professor, Fashion Institute of Technology
Other changes are operational. The company recently began allowing consumers to shop the site directly, an option it calls "Direct Buy," rendering it a less differentiated apparel e-retailer. And for weeks now the company has been dealing with fallout from new stylist work restrictions, prompting concern from some analysts and, eventually, an apology from Chief Operations Officer Minesh Shah.
The strains are catching investors' attention. Wells Fargo analysts last month warned that the company didn't benefit that much from the pandemic-related rise in e-commerce, but also may not be helped much now by the late-pandemic recovery.
Is it the algorithm?
One area of tension at Stitch Fix centers on how to balance the contributions of its algorithms and its human stylists. Company executives and spokespeople have long touted both.
In his internal memo to stylists last month, Shah emphasized their importance, though about a dozen stylists, sharing their perspectives by phone, email and text, say they remain skeptical.
"Our highly skilled, expert team of stylists have been a differentiator for Stitch Fix for more than a decade, and that is not changing," Shah wrote, according to copies of the memo sent to Retail Dive and verified by the company. "In the future, we believe there will be even more opportunities for stylists to connect with and deepen their one-to-one relationships with clients through services and skills we know only our stylists can provide — coaching, advice, inspiration."
The process does allow for the human touch. Stylists order the items for each fix. Customers can write notes to them about how happy or disappointed they may have been, to tweak their preferences or let them know they don't want jeans or have a wedding coming up — and stylists write back. Many customers keep mood boards with pictures or Pinterest posts to demonstrate their taste. One-on-one live styling by video may be on the horizon.
But several former and current Stitch Fix stylists, in conversations with Retail Dive in recent weeks, said they are convinced the company is increasingly marginalizing the stylists' role in favor of its tech. (All spoke on condition of anonymity out of concern for their jobs or exit payments.)
In a process still being tested, the algorithm is now often responsible for choosing some if not all of the 10 items customers are shown in their Fix Previews, for example. When it picks things a customer has said they never want, overloads a preview with duplicates or accessories, or suggests garments out of season, customers are sometimes angry with their stylists for not listening to them, stylists said. The tech seems daunted by natural language, so will choose "jeans" or "blue" when a customer says "no jeans" or "nothing blue."
But stylists also said they aren't permitted to explicitly call out the algorithm as responsible for such mistakes. Yet transparency around this could alleviate a lot of frustration because people have different expectations of humans versus bots, according to Jonathan Zhang, a Colorado State Business School professor whose specialties include consumer behavior, data analytics and machine learning and who has followed Stitch Fix for most of its tenure. It would be more fair to the customer and the stylist for Stitch Fix to flag whether its algorithm or a human is responsible for what a customer sees, but the company may be overconfident in its technical prowess, he said.
A Stitch Fix spokesperson declined to address whether the company is aware of tech glitches in absorbing client feedback or how transparent the company is when the algorithm is in charge.
"Stitch Fix is part of a generation of companies that promise consumers, that, with better machine learning, with better data and better AI, we can do magic for you, right?" Zhang said by phone. "But oftentimes they fall short of the promise. There is almost like a technical chauvinism towards AI and the promises of AI."
In reality, the algorithm has "far to go," according to a stylist still employed there. "I was shocked that they are leaning so strongly on the algorithm right now because it is not ready," the stylist said by email. "It doesn't recognize seasonality which is part of the reason your Fix Preview has a bunch of sweaters in the middle of the summer."
Disruption in fashion
Over-reliance on algorithms can be a problem not just in sales and marketing but also for merchandising, because the tech can't fathom what's going on without stylists' input from the field, according to FIT's Carter.
"It goes back to the same problem we've always had in retailing — the operations side of the business does not talk enough with the merchant side of the business," Carter said. "Think of the stylists as the merchants, the AI people are your geeks. A computer can be designed to be informed enough to give you rationale, but you're going to need the human factor. If these geeks who do the algorithms don't interface with the stylists, they will not know whether, for example, your high return rate is because of poor fit, or because the quality was not the same as the expectation from the information or the picture you put online, or because customers decided it's no longer in fashion."
Because Stitch Fix has thousands of stylists, their impressions would need to be collected systematically, probably by region, but regularly — once a week or at least once a month, she also said. "This is basic retail management, inventory management, but you have to have the merchants, the analysts, the algorithm engineers, all communicating as a team," she said. "The algorithm gives you the 'what,' but the 'why' is just as important. What they do at Stitch Fix and in too many companies is keep them separate like church and state. They need to be talking to each other on a regular basis."
Stylists do have a number of ways to communicate about the merchandise, according to a company spokesperson. After they style a fix, for example, they're asked whether they had appropriate inventory for the client. There's also an open internal forum to share feedback and solve problems, and twice-yearly in-depth surveys. Leads, who manage stylist teams, regularly meet with merchandising teams, the spokesperson said.
But stylists say that communications are mostly a one-way street, designed to inform stylists of fashion trends and inventory levels rather than to get their input. The post-fix report entails marking it "easy, typical or difficult," a stylist said by email. "If difficult, we can select from additional reasons but we can't make inventory recommendations. We simply check the box that says 'couldn't meet client request.' I've never heard how this information is used."
Stylists also described communication with leads regarding merchandise as a loose process. "They may ask for feedback, but nothing changes," one stylist said by text. "I've seen the same inventory the past 3 years I've been with them. When we escalate letting them know the inventory is challenging, nothing happens. No feedback email or anything. It just goes off somewhere in a vacuum."
The surveys are inconsistent and seem like a black hole, stylists also said. "The last one I received has nothing to do with merchandise and all to do with employee satisfaction," one said. "It's clear they don't take these into consideration. Sending them is simply performative."
Available inventory is a major source of frustration for Stitch Fix stylists these days, and many told Retail Dive that's leading to fixes that include unseasonable or otherwise unwanted items. Children's boxes were especially difficult to fill lately because those require 10 items and there has been a dearth of inventory for back-to-school, according to one stylist.
"They release quarterly updates but I almost never see the inventory they say we're getting," a stylist said. Another said that when she notes that inventory has been challenging, she's told to "get creative."
Some of that is likely down to a supply chain still in turmoil, Zhang said. Inventory issues related to the pandemic began surfacing last year as the disease outbreak roiled manufacturing, shipping, warehousing and fulfillment, and the problems seem to be only worsening this year.
"Some problems are beyond their control," he said. "These are more macro issues, I mean, there is nothing that they can do, right? And then there's another set of problems that may be unique to Stitch Fix."
That includes the costs of selling clothing online plus added limitations introduced by the box subscription model. For example, returns, already rampant in apparel e-commerce, are built into Stitch Fix's "try before you buy" approach. Plus, in order to most efficiently and accurately fill one box with five items for a given customer, stylists must choose from the inventory at a single warehouse that the algorithm points to as best fulfilling their style objectives, a company spokesperson confirmed by email.
That's become a thorny issue for a couple of reasons, according to stylists. For one, the lean inventory is making it difficult to fill fixes that meet customers' parameters. For another, now that customers can view Stitch Fix's entire inventory via Direct Buy, some are irritated that certain items aren't available through their fix. Customers often want to include an item they've found themselves on the site in order to boost their chances of keeping something and not wasting that $20 styling fee, stylists said.
Similarly, Stitch Fix often depicts garments in social media posts or its Style Shuffle app that it doesn't actually sell, or perhaps was only available in small quantities or to customers who chose the most luxe budget in their initial quiz, according to several stylists. Some also say they feel set up for failure when the company tells a customer on social media that their stylist can help them find an item they've admired on a post. A Stitch Fix spokesperson didn't address this directly but said the company does sometimes leverage Style Shuffle to test the response to private label items before they're manufactured.
The real disruption?
While Stitch Fix doesn't call itself a subscription, preferring the term "styling service," many customers and analysts perceive it as one.
In the last three years, U.S. consumers increased their spending on subscriptions by 15%, were less aware of that spending and tried new such services, according to recent research from consulting firm West Monroe. But apparel boxes like Stitch Fix and Nordstrom's Trunk Club have proven to be less sticky than other consumer product and service subscriptions, according to that report. There are many subtle, changeable yet significant variables in apparel choices — including fit, texture, color, price, occasion, seasonality, age, taste, fashion, preferences — that could spur returns or, over time, cancellations, according to Dhaval Moogimane, a West Monroe partner and report co-author. The loss of the styling fee and the need to repackage unwanted items may also be discouraging, he said.
"It's a little bit more capricious," he said by phone. "But I also think, probably, it creates an opportunity for them to be able to leverage the designers because not everybody is on top of fashion. The value proposition could be 'We are kind of pushing you towards where fashion is going, and there's some experimentation of this process.' There's a way to sell that as a positive, it's just a little bit more complex."
An idea that seems simple on paper may not be so in practice. Stitch Fix has arguably dedicated 10 years, an intricate and constantly evolving algorithm and a team of thousands of stylists to essentially replace what most of us often do with relative ease (mull what to wear), relative speed (browse online or in a store) or relative enthusiasm (shop for something new). Fit, budget, style-fabric-color-pattern preferences and all the other criteria either don't need to be articulated or can be discussed casually, using natural language, with a store associate. It's somewhat like the last mile, which was long accomplished without much thought by store customers — picking items off the shelves and driving the purchases home themselves — but is now confounding retailers that sell online.
Ultimately, the reason Stitch Fix isn't all that disruptive is because its addressable market is pretty slim, according to Forrester Principal Analyst Brendan Witcher. Their core customer is someone who cares about fashion enough to want nice clothes, but not enough to explore that on their own.
"That's a very narrow band," he said. "If I care so much about food or about clothing, then I'm not going to use a Blue Apron or a Stitch Fix, because I want to do that myself — part of my joy is buying food or buying clothes. If Stitch Fix has been disruptive for any reason, it's because their approach is, 'We're not trying to be on trend or trend-setting, we're just trying to make people feel good in clothing. Now the question is — are they being successful? How do you interpret a person's fashion the way they interpret it? Our bodies are just so unique. The way fabric feels, the cut of a sleeve. We're all just so unique and so different, and our likes and dislikes are very hard to determine."
Editor's note: This story was first published in our weekly newsletter, Retail Dive: Operations. Sign up here.