Case study • One Skin

Challenge

OneSkin’s growth was outpacing the support team in the worst way. First response time had stretched from 15 minutes to 10.5 hours. Every CX manager knows what that number does to a brand.. Cancel-with-a-question turns into cancel-and-leave. Subscriber LTV erodes one un-replied ticket at a time.

The team did the math the way every operator does. Closing the FRT gap by hiring meant adding additional agents, plus management overhead, plus the recurring training cost, plus the fact that subscription volume was still climbing and a Headcount wasn’t a strategy. It was a stopgap that would arrive late, cost too much, and not solve the underlying problem.

The ticket queue was dominated by subscription management:

• Cancellations, frequency changes, skip requests, order modifications
• Every action requiring an agent to click through Skio, Shopify, and Order Desk in sequence
• Email handling all the largest-channel volume manually
• A pending Zendesk-to-Kustomer migration that couldn’t share the runway with a slow rollout

Leadership set a clear bar for any AI investment: 20 to 40% ticket reduction, with a clean ROI story, and the deployment had to be live and proven before the platform migration so it wouldn’t compound the switch risk.

Solution

OneSkin runs KODIF across chat and email, with native integrations into Skio, Shopify, and Zendesk. The deployment started narrow on purpose: WISMO and subscription cancellation first, then expanded into skip-next-shipment, returns, and order cancellation as the team built confidence in the policy logic.

Phase 1

Subscription cancellation, with a retention conversation built in

• The AI Agent runs the same conversation a trained agent would: identify the reason, offer the right retention option (15% discount when it’s price-driven, frequency change when it’s cadence, pause when it’s “I have too much product”)
• Cancellation only processes if the customer confirms after the offer
• Real-time write-back to Skio

Phase 2

WISMO and lifecycle policies across chat and email

• Real-time order tracking pulled from Shopify and business-day delivery calculations
• Skip-next-shipment, letting subscribers defer a delivery without an agent in the loop
• Returns initiate on chat with a Zendesk ticket created automatically and full conversation context attached
• Email policy runs the same logic as chat, with deliberately conservative routing on refunds and complex billing

Results

Across 2,385 eligible chat conversations, KODIF contained 75%, with 52% fully automated and no human handoff. Subscription cancellation specifically runs at 4.67-star CSAT, with all Skio-connected tools sitting between 4.67 and 4.75 stars. Customers who can self-service their subscription lifecycle are highly satisfied with the experience, even when the outcome of that conversation is a cancellation.

• 589 chat policy executions (the action-taking work: cancellations, WISMO, skips, returns)
• 1,153 email conversations resolved autonomously: 433 policy executions plus 653 FAQ responses
• 60% of CSAT responses landing at five stars
• 86.4% chat engagement rate, meaning customers actively conversing past the first prompt

Policy executions are growing faster than total volume, which means the AI is taking on more action-requiring work per conversation as it matures, not just deflecting more of the easy ones. That’s the curve a CX leader actually wants to see.

Crucially, none of this came at the team’s expense. KODIF and OneSkin’s CX headcount are complementary, not competing: the platform absorbs the repetitive, policy-driven volume — cancellations, WISMO, skips, returns – so the human team can stay lean and spend its time on the judgment-heavy work that actually needs a person. Headcount was never the enemy. KODIF simply lets OneSkin run a more streamlined team at a lower cost than scaling the org chart would have required.

What's next

OneSkin is extending the deployment in five directions:

1. 25–30% deflection rate target for 2026
2. Order cancellation policy (Shopify + Order Desk integration)
3. Zendesk → Kustomer migration + AI Policy migration
4. Pre-sales concierge for 2027 retail launch
5. Frustration Index smart tag for advanced sentiment tracking