Skip to main content
Apolla AI
A personal care brand

Consumer goods

First-contact resolution up from 54% to 81%, transfers to human agents halved

Personal care brand: a consumer service Agent measured on resolution rate, not response rate

Challenge

Hundreds of thousands of inquiries and complaints arrived across channels every month. The old bot matched keywords, and consumers were handed to a human within three messages. Staff turnover was high, product and returns-policy training could not keep up, and the complaint escalation rate stayed stubbornly high.

The brand wanted to move service from answering fast to actually resolving, and to feed inquiry data back into product and marketing.

Approach

We rebuilt the service Agent around a single core metric, resolution rate, covering product questions, order status, returns and complaint intake, working alongside human agents.

  • One unified product and policy knowledge base; the Agent cites its sources, and policy changes take effect the same day
  • Order lookups, return requests and other transactional flows are completed by the Agent itself, with no handoff
  • Complaints are triaged by sentiment and risk; high-risk cases go straight to senior staff with full context
  • Unresolved issues are clustered weekly and sent to the product and marketing teams

Architecture highlights

Integrated with order, membership and logistics systems for transactional tasks. Every conversation enters the evaluation set; an offline evaluation runs every two weeks, and any intent whose resolution rate falls below threshold goes automatically into the optimization queue.

Customer feedback

A fast bot reply means nothing. Consumers want the thing done. Now eight in ten are done in the first round.
Head of Consumer ServiceA personal care brand

Want more detail on this project?

In a workshop we can share the architecture decisions, the pitfalls and the actual timeline.