Dimension Labs/Live Monitor
Refreshes weekly · Lemonade omni-channel

Omni-Channel Voice Monitor

Lemonade · week of May 12, 2026 · 4 review surfaces + 623 conversation sessions · the omni-channel metrics from the Causal Brief, tracked live
4 metrics need attention
Dead-end rate
18%
rising
of “contained” chats were actually dead-ends — the customer gave up or never got the human they asked for
Hidden inside the 88% containment rate. Target under 5%.
Escalation gap
61%
rising
of customers who clearly needed a person never reached one, in the chat logs
The reviews only inferred this. Target 25%.
Genuine resolution
70%
steady
of chats were genuinely resolved — against the 88% the bot reports as “contained”
The honest containment number.
Chat → 1-star risk
82%
steady
of failed or abandoned chats already carry a one-star review signal
The leading indicator of the public score.
One-star share
31%
rising
of all public reviews this period
Target under 25%.
AI-failure mentions
11%
rising
of reviews name a specific AI failure; “refused to escalate” leads
Was ~7% early 2025.

What to watch: containment vs genuine resolution

The bot “contains” 88% of chats, but only 70% are genuinely resolved. The shaded gap is the dead-ends — customers who gave up or never reached a human, and the leading edge of the one-star tail.
60%80%100% 88% “contained” 70% resolved the 18-pt dead-end gap Jan’25AprJulOctJan’26Apr
18%
dead-end gap now
15%
12-month average
9%
best month
<5%
target

Containment on its own says everything is fine. The honest number is the resolution line, and the gap between them is your earliest warning: when it widens, the one-star tail grows the next week. Watch the gap, not the headline.

Alerts this week

Containment is crediting failure: 88% reported, 70% real.
Nearly one in five “contained” chats is a dead end — the customer gave up or never got the human they asked for. The metric the team optimizes is rewarding the worst outcome.
→ Split containment into resolved vs dead-ended on the dashboard, today.
Escalation gap is 61% in the actual chat logs.
Six in ten customers who asked for a person never reached one — far worse than the reviews could show, and the single strongest driver of one-star reviews.
→ Audit the human-handoff path on disputed claims and bot loops.
82% of failed chats are one-star reviews in waiting.
Failed and abandoned conversations carry a would-1-star signal at 82%. The chat is the earliest copy of the public score — weeks before the app store, months before the regulator.
→ Treat failed-escalation chats as the live early-warning queue.
What customers ask the AI — and where it dead-ends

The bot nails transactions and dead-ends on the moments that decide loyalty.

Every conversation classified by what the customer came for. The four transactional intents resolve cleanly; the relational ones — disputes, cancellations, claim-status chasing — are where the bot dead-ends and a human is needed.
WHAT CUSTOMERS BRING TO THE AI 623 sessions, by intent · bar length = volume, red = dead-ended New quote 150 File a claim (FNOL) 110 Billing & payment 104 Coverage question 89 15% dead-end Complaint / dispute 70 100% dead-end Claim status chase 52 35% dead-end Renewal 35 Cancellation 13 100% dead-end Handled by the bot Dead-ended / needed a human

Quotes, first claim notices, billing, and renewals are handled end to end. But 100% of complaints and cancellations and 35% of claim-status chases dead-end — the exact moments that decide whether a customer stays. The content of the queue tells you precisely where to place the human: not everywhere, just on the relational tail.

Recommended actions this week
ACTION 01

Route the relational intents straight to a person.

Complaints, disputes, cancellations, and stalled claim-status chats dead-end on the bot. Auto-route these intents to a human on first detection — they are where loyalty is won or lost.

Moves: escalation gap, dead-end rate
ACTION 02

Stop trusting containment. Split it.

Report “genuinely resolved” apart from “no human handoff,” and alarm on the gap. The dead-end rate, not the containment rate, is the number that predicts the one-star.

Moves: dead-end rate, genuine resolution
ACTION 03

Watch the conversation signals as early warning.

Failed escalations, loops, and “no one to call” predict the one-star weeks ahead. Stand them up as live signals; 82% of failed chats are one-stars in waiting.

Moves: chat → 1-star risk, one-star share

On the watchlist

Dimension Labs · Omni-Channel Voice Monitor · review + conversation figures, refreshed weeklyGenerated June 2026