Skip to main content
Case studies
Linen HouseHome goods · EU
82%of CS auto-resolved

Linen House handled a growing support queue with four agents and a shared inbox. Today one human oversees a CS Specialist that resolves 82% of tickets — and the remaining 18% is the work humans should actually be doing.

// the deltas

Tickets auto-resolved

82%
+82pp

Median first reply

3m
-41m

CS headcount

1
-3

Voice fidelity floor

0.78
gated

The queue was mostly the same five questions

Like most home-goods stores, Linen House's support volume was dominated by a handful of repeatable cases: where's my order, can I return this, can I swap the size. Four agents spent their days on work that was important to customers but rote to perform.

The risk in automating it wasn't accuracy — it was voice. A refund approved in a tone that didn't sound like Linen House would do more damage than a slow reply.

Policy-grounded answers, voice-gated replies

The CS Specialist resolves refund-eligible orders, shipping queries, and policy-bounded exchanges end-to-end, grounded in Linen House's own policy chunks. Every drafted reply passes through Reply Judge, which scores it against the brand-voice anchor built from the team's best 200 sent replies.

Anything below 0.78 fidelity doesn't send — it lands in a human's inbox with the score and the lines that pulled it down. That 18% is now the entire job of the one remaining CS lead.

Half the noise, none of the drift

Auto-resolution hit 82% within the first month and held. Median first reply dropped from 44 minutes to 3. But the quieter win is consistency: the brand-voice anchor became a reference the whole team points to, so even human replies got more on-voice, not less.

We didn't fire three people — we moved them to the work that needed a human. The agent does the refund-eligible, policy-bounded 82%. Reply Judge makes sure nothing goes out that doesn't sound like us.
Maya Chen
Head of Ecommerce · Linen House

// your story next

See your own case study in 90 days.

The fastest path to a numbers-driven story is to connect a read-only token. We'll show you exactly what the agents would have done over your last 30 days — before anything writes.

Dry-run by default · Append-only logs · One-click rollback