333. Why 32 Degrees Won't Celebrate 80% Automation Alone (and the Return Data That Caught a Manufacturing Error) - Jennifer Villalba, 32 Degrees

Jennifer Villalba's team found a real manufacturing fault in a new pair of jeans not through a lab test, but through a spike in exchange requests and a few honest lines in a returns form. That is the kind of thing that happens when a CX team is treated as more than a place tickets go to die.

Villalba is CX director at 32 Degrees, the apparel brand, where she has worked for almost four years. She started her career in customer experience five years earlier at Apple's retail stores, and has spent close to a decade in CX overall.

CX as a revenue team, not a cost centre

Villalba says CX has long been treated as a cost centre, a place that answers questions rather than builds the business. She and other CX leaders are working to change that, proving the function can drive revenue too, not just handle complaints. Done carefully, even a transactional post purchase email can carry a relevant product suggestion, as long as it leads with the value the customer actually wants first.

The tracking page nobody expected to win

One of the biggest changes this year has been moving post purchase notifications off Shopify and onto ParcelLab. The order tracking page it powers has become the most visited page on 32degrees.com, drawing 200,000 sessions in a single month. Villalba's manager, the head of e-commerce, told her the tool had already paid for itself from that traffic alone, even though she had brought it on for CX reasons under her own budget.

When return data catches what a lab test would miss

When 32 Degrees launched its first pair of jeans, Villalba noticed an unusually high exchange rate. Digging into return reason codes and reviews through Loop led her back to a sizing guide error from the manufacturer. The jeans were pulled from the site, fixed and relisted before the issue escalated with customers.

"It would have never been, if we weren't seeing comments from customers or the data from returns, they probably would have eventually found out, but we were able to get a hold of it in the beginning before it went crazy and wild."
Jennifer Villalba

A hoodie complaint becomes a product review

A similar pattern is playing out with a lightweight hoodie. Customers keep saying it doesn't feel like the cozy, wintery hoodie they expected, even though nothing is technically wrong with it. Villalba suspects the issue may be positioning and description rather than the product itself, and it has been flagged to the buying and production teams ahead of next year's range.

Filling the hours human agents can't

32 Degrees only staffs live support from 9am to 5pm Eastern, with agents working until 10pm, but customers still reach out overnight wanting to edit an order, change an address or cancel entirely. Villalba is testing an AI chat bot to cover exactly those simple, well defined requests around the clock, rather than waiting for a slower, safer rollout after peak season.

80% automation isn't the win

Villalba is wary of chasing automation percentages for their own sake.

"AI is great, but I also want to be really intentional with it, and think about what the gaps are in our business. Okay, we have 80% automation. But were those conversations great, or was the customer frustrated?"
Jennifer Villalba

Will Laurenson made a similar point about capacity rather than headline numbers: "If all the quick and easy inquiries are dealt with by AI, then anything that does come through to the team, the team knows that there's a reason it's coming through, and so someone can jump on it."

Peak season prep and split shipments

32 Degrees does not run one single site-wide Black Friday event. It runs weekly flash sales year round, so a discount code is one part of a much longer preparation process that starts as early as July or August. That includes auditing SOPs, FAQs, policies and product pages, and setting clear KPIs like a six hour first response time target.

This year also brought two new complications: a new outsourced support team onboarded weeks before peak, and a new heated apparel line with split shipments, where the battery ships separately from the vest or jacket. Both meant new training and new customer messaging had to be built at the busiest possible time.

Key takeaways

• Treat returns data and review comments as an early warning system for manufacturing and sizing problems, not just a refunds process
• A product complaint may be a positioning or description problem rather than a product problem, so check both before changing the product
• Proactive shipping updates reduce support volume rather than increasing it, because silence is what drives customers to reach out
• Automate the simple, well defined requests around the clock so human agents can focus on conversations that need a person
• Judge automation by whether the conversation actually solved the problem, not by the automation percentage alone
• Start peak season prep months ahead of any new complexity, like a new outsourced team or a new product with split shipments

‍

Villalba says the goal is simple: let customers solve their own problem when they can, and make sure every conversation that does reach a human agent is one that is actually worth their time. Listen to the full episode to hear how she is getting 32 Degrees ready for Q4.