Copperline Outfitters sells tents, packs, technical apparel, and camp kitchen gear directly to customers from its Denver headquarters. About 70% of its revenue comes through its Shopify storefront, and nearly a third of the year's orders land between mid-November and the end of December.
That concentration made customer support a seasonal hiring problem. In 2025, Copperline solved it a different way.
The challenge
Copperline's year-round support team is 38 agents. In a normal week they handle about 14,000 conversations across email, chat, and SMS. From mid-October through December, volume runs at more than three times that.
For five years, the answer was temporary hires. Each October, Copperline recruited around 110 seasonal agents through a staffing agency. Each one needed three weeks of training on returns rules, warranty terms, and the product catalog. By the time they were fully productive, peak was half over. By January they were gone, along with everything they had learned.
The costs added up:
- Around $1.6 million a season in agency fees, wages, and training time.
- Six weeks of senior agent time spent on training and shadowing.
- Quality dips every December. QA scores dropped an average of 11 points during peak, and first-response times on email stretched past 30 hours.
The contact mix was predictable, which made it more frustrating. More than half of peak volume came from five topics: order status, order changes, returns and exchanges, damaged items, and promo codes that didn't apply.
"We weren't short on knowledge. We were short on hands, for six weeks a year, on questions we'd answered a hundred thousand times." — Sam Duarte, Head of CX Operations
Why Pazau
Copperline had tried a chatbot in 2023. It deflected questions to help center articles and pushed customers toward a contact form. Customers hated it and CSAT fell. The team turned it off after four months.
Sam's requirement for the next attempt was simple. The AI had to finish the job. If a customer wants to exchange a medium jacket for a large, the AI should create the exchange, not link to the returns page.
Pazau Agent met that bar. It integrates with Shopify for orders and Loop for returns. It can take real actions with limits that Copperline's CX team sets. The team could also watch every decision through the trace, which mattered to a team that had been burned before.
Pricing mattered too. At $0.75 per AI resolution on the Business plan, Copperline's finance team could model peak season cost directly against the agency invoice.
The rollout
Copperline signed in April 2025, which gave the team seven months before Black Friday.
May: Inbox first. The team moved from Gorgias to Pazau Inbox so AI and human agents would share one queue. Macros, tags, and SLA rules came across in the first two weeks.
June–July: Procedures and Actions. Copperline connected seven Actions: get_order, update_shipping_address, cancel_order, create_exchange, create_return_label, issue_refund, and apply_discount. Priya Nair, the team's senior QA lead, wrote the first Procedures with Sam:
- "Where is my order" checks fulfillment and carrier tracking, and explains delays in plain language.
- "Change or cancel before shipment" edits the address or cancels the order if it hasn't left the warehouse.
- "Size exchange" creates the exchange in Loop and emails a prepaid label.
- "Damaged item → Replace or refund" asks for a photo, then offers a replacement or refund. Refunds over $250 need a CX lead's approval.
- "Promo code didn't apply" checks eligibility and applies the discount retroactively when the code was valid.
- "Warranty claim" collects photos and the product SKU, checks the lifetime warranty terms, and routes approved claims to the repair team.
August–September: Simulate and tune. Every Procedure ran against the previous year's peak conversations. Simulation found a gap. Gift orders were being refunded to the purchaser when the recipient asked for an exchange. Priya added a gift-receipt step in an afternoon.
October: Soft launch. Pazau Agent went live on chat, then email, six weeks before Black Friday. By November 1 it was resolving 64% of conversations.
"Simulation changed how we write policy. We used to find out a rule was wrong from an angry customer. Now we find out from 2,000 past conversations before anything goes live." — Priya Nair, Senior QA Lead
The results
Black Friday–Cyber Monday 2025 was Copperline's largest peak ever. Order volume was up 22% on 2024. Over the ten-week peak, the team handled 486,000 conversations, 3.4 times normal volume.
Copperline hired zero seasonal agents, down from 110 the year before.
- 71% AI resolution rate across peak season, rising to 76% on order-status questions.
- Median first response of 6 seconds on chat and 2 minutes on email, compared with over 30 hours on email the previous December.
- QA scores held steady. Pazau Insights scored 100% of conversations. The team's average stayed at 93, with no December dip.
- Roughly $1.6 million in seasonal staffing costs avoided. AI resolution fees for the same period came to about a sixth of that.
- CSAT rose 5 points during peak, from 84 to 89.
The year-round team spent peak on the conversations where people help most: gear advice before a big trip, complex warranty cases, and orders that went wrong in ways no Procedure anticipated.
Insights also caught something the team would have missed. Damaged-on-arrival contacts rose 41% in the second week of December. The spike traced to one fulfillment partner and one packaging change. Copperline fixed it before the holiday rush peaked.
What's next
Copperline is expanding Pazau Agent to its B2B channel, where outfitters and guide companies place bulk orders. The team is also connecting its inventory system so the agent can offer back-in-stock alternatives when an exchange size is sold out.
Peak season 2026 planning, Sam says, took one meeting instead of three months. The plan is the same team, the same Procedures, and more volume.