Book A Demo

Case study • Get Joy

When Get Joy went national on Chewy, their two-person
support team needed a policy layer that could scale
without adding headcount

  • 89%

    Chat containment rate

  • 48%

    Conversations resulting in Automation

  • 4.77★

    Avg CSAT, 66% positive ratings

Challenge

Get Joy’s support team was built for roughly 800 contacts a month. National distribution through
Chewy and Phillips changed that number fast. Volume projections pointed toward 5,000+
monthly contacts, and the existing model, two in-house agents plus a Talent Pop outsourcing
contract wasn’t sized for it.

Zendesk’s native AI handled surface-level FAQ deflection but couldn’t execute the policy depth
Get Joy needed: checking Skio subscription records by email, modifying billing dates, running a
cancellation flow with a retention offer built in. Those conversations kept landing on humans. A
parallel problem sat entirely outside the support stack: veterinary nutrition guidance was tracked
in a Google Sheet, disconnected from everything else.

The team needed an automation layer that could execute policy, not just deflect FAQs. The
evaluation came down to a question of policy depth: could the tool handle the conditional logic
that subscription management actually requires?

  • WISMO contacts rising as they added new subscribers across the country
  • Subscription management (skip, reschedule, cancel, modify) requiring agents to
    navigate Skio, Shopify, and Zendesk in sequence
  •  Recharge-to-Skio migration underway, with the support tool needing to move with it
  • Escalation paths that dropped every unresolved conversation directly on an agent, with
    no availability check or context pre-populationWISMO contacts rising as they added new subscribers across the country

The team needed an automation layer that could execute policy, not just deflect FAQs. The
evaluation came down to a question of policy depth: could the tool handle the conditional logic
that subscription management actually requires?

Solution

Get Joy runs KODIF’s chat agent, which they appropriately called JOY, with direct integrations
into Shopify, Skio, and Zendesk. The configuration covers three core contact drivers: order
inquiry, subscription management, and escalation routing

Phase 1: Order lookup and WISMO

JOY pulls order status directly from Shopify by email or order name. Customers asking about
shipment timing or delivery details get the answer in the conversation.

  •  Shopify order lookup by email
  • Shopify order lookup by order name
  • Covers shipping status, estimated delivery, and last tracking even

Phase 2: Subscription management on Skio

The subscription workflow covers the full account lifecycle: lookup, review, modification, and
cancellation with a retention step. JOY executes changes directly in Skio rather than creating a
ticket for a human to action later.

  •  Customer and subscription lookup by email
  • Billing date and frequency modifications
  • Subscription cancellation flows with a retention offer before confirmation
  • Legacy Recharge handling for subscribers mid-migration

Phase 3: Escalation architecture and Zendesk integration

When a conversation needs a human, JOY checks agent availability first and creates a Zendesk
ticket with context already populated rather than handing off a blank conversation.

  •  Zendesk agent availability check before any escalation decision
  • Zendesk custom ticket creation with conversation context
  • Live agent handoff with structured context transfer
  • SMS channel running the same policy logic as chat

Results

89%

Chat containment

16%

Cancellation saved

4.77

CSAT

Across all conversations, KODIF contained 89%.

Automation:

  •  48% of conversations used automation executions, covering order inquiry, subscription
    management, and cancellation
  • Order Inquiry: the highest-volume policy, reflecting active subscriber tracking of fresh
    meal delivery windows
  • Subscription management and cancellation: 16% of conversations executed cancellation
    flows including a built-in retention offer step
  • 13% explicit handoff rate
  •  CSAT across rated conversations returned a 4.77-star average, with 66% of responses
    at 4 or 5 stars.

What's Next

Get Joy’s next build phase covers five directions:

  1. 1

    a promotions and discount lookup policy to surface active email offers within the chat flow

  2. 2

    a proactive subscription alert workflow for billing and shipment changes

  3. 3

    completion of the Recharge-to-Skio migration to simplify the subscription tool path;

  4. 4

    vet nutrition guidance policies to replace the Google Sheet-based workflow

  5. 5

    and SMS policy parity to match full chat coverage as subscriber volume grows