For Cathy Polinsky, the position of chief technology officer (CTO) at Stitch Fix Inc. was a perfect fit.

She said she hates shopping for clothes and applauds the Stitch Fix online business model of having personal stylists — backed by data and style-matching algorithms — select items customers might love and mail to them in a box.

Plus, Polinsky was destined to code. She started programming with the Logo language in first grade on the Apple II, and became enamored with computer science in high school and college. She emerged as a software engineer, first for a dot-com startup and later at Amazon.com and Salesforce.com.

Stitch Fix announced Polinsky's addition as CTO in November 2016, in charge of IT, engineering, security and user experience (UX).

"What really attracted me to Stitch Fix is that I'm fascinated by growth, and Stitch Fix is in this amazing spot that's disrupting a whole industry," Polinsky said in an interview with CIO Dive. (CIO Dive sister publication Retail Dive named Katrina Lake, founder and CEO of Stitch Fix, 2018's Disruptor of the Year.)

Now, nearly two and a half years later, Polinsky said she's proud of her team's accomplishments but there's still room for tech-enabled growth: "I feel like we have more opportunities today than we did when I walked in the door at Stitch Fix."

A team for a burgeoning business

When Polinsky started, the technology team had about 60 software developers and UX engineers. The first step was doubling the size of her team so it could support, rather than constrain, the burgeoning business.

"The company had grown really fast around us, but the technology team hadn't kept pace with the rest of the business, so the team was a constant bottleneck for projects and really felt like they could never keep up," she said.

The larger engineering division has helped Stitch Fix expand into menswear, plus sizes, "extras" (such as socks and underwear), and kids. For each new market launch, the technology team has to update the inventory, merchandising, styling, fulfillment, and customer-facing applications to support the special characteristics of that market.

The launch of Stitch Fix Kids last summer required connecting with a new category of clothing vendors, and paying extra-careful attention to clothing sizes because children are "always changing," Polinsky said. "They're also not signing up for themselves, and so we had to think about the experience of a parent signing up for their kid, asking questions they may not know how to answer unless their kid is sitting right next to them."

Polinsky's technology group is also heavily involved in supporting international expansion — starting with a launch in the U.K. later this year. "That has been a really exciting project for us," she said, "dealing with international currency, language, and new shipping modes, as well as GDPR."

Adapting systems to Europe's General Data Protection Regulation has been a "heavy lift," Polinsky said, but is consistent with the company's efforts to ensure its systems comply with various government regulations, including the Sarbanes-Oxley Act and California's privacy law.

"We're a new public company, so in IT we've done a lot to add controls and support for running a mature public company," she said.

Stitch Fix went public in November 2017 as a Wall Street darling, but disappointing earnings in the second half of 2018 made some analysts wonder whether the "fix," as the delivered box is called, was a passing fad.

However, Stitch Fix reported net revenue up 25% from last year in the second quarter of 2019.

A key metric: The number of active clients increased 18% to 3 million — and on average they each spent 6% more than the previous year.

Counter-intuitive UX

The technology group constantly tests ways of improving the onboarding process for new clients, Polinsky said. For most consumer websites, this would mean removing steps in the signup process, but that's not necessarily true for the Stitch Fix model.

Polinsky's team has found through A/B testing that removing questions — for example, whether a customer likes or hates polka dots — from the lengthy style questionnaire actually reduces signups. That was a head-scratcher until they surmised that customers are willing to answer numerous questions to gain "confidence that we know them well enough to style them well," she said.

At Stitch Fix, Polinsky's group works very closely with the separate data science team, where over 100 data scientists handle not only the style-matching algorithms but also the algorithms for marketing and for optimizing back-office operations.

"We are one of the most data-centric companies that I've ever worked in, and we use experimentation and data science for everything that we do," Polinsky said.

Thumbs-up or thumbs-down

But algorithms work best with lots of data.

Stitch Fix collects more than 90 data points in the initial style profile, and also gets signals from what clients keep or reject on the back end.

To gather even more clues about customer likes and dislikes, Stitch Fix developed an application called Style Shuffle. The Tinder-like game shows a series of clothing items and lets clients give each one either a thumbs-up or a thumbs-down — thus providing more data to feed the style-matching algorithm.

"Clients love it. It's super engaging. It's fun. We have over a billion ratings now on this platform. Over 75% of our active clients have tried it at least once. Clients who have played with Style Shuffle have better 'keep rates' — they're keeping more items because we're sending more relevant items and really getting their style," Polinsky said.

What happens if a client has a low keep rate?

Stitch Fix has a client experience team that will reach out to the client and ask them to play Shuffle; seek answers to more style questions; and see if the client wants to work with a different stylist, Polinsky said.

"We want to make sure that we've got the right signals coming from the algorithm, and that we've got the right person that gets to know you and styles you well," she said. "This is a service that gets better and better over time. The more we get to know you, the better we can really get you."