Added
- Customer risk score — every customer gets a 0–100 risk score, recomputed nightly from three signals: how their recent messages feel (with this week's mood counting more than last month's), how many of their threads escalated to P1/P2, and how many negative messages they've sent in the last 30 days. A chatty-but-happy customer never scores as risky — volume alone isn't a signal.
- Slack early warning — when a customer crosses the risk threshold with repeated negative messages, ReplyMill posts one alert to your Slack channel with the specifics: "3 negative messages, 2 escalated threads." One grumpy email never pages you, and the same customer won't re-alert for a week.
- At Risk Customers on your dashboard — the riskiest customers, with their name, signals, and score, each linking straight to their latest thread.
- At-risk badges in the inbox — threads from flagged customers get a high-contrast "At risk" chip in the thread list, plus a Health row in the conversation details panel.
- Sentiment on every message — ReplyMill now reads the mood of every customer reply, not just the first message of a thread. A customer who starts polite and turns frustrated mid-conversation shows up in the score.
Why it matters
Churn rarely announces itself. It looks like a customer whose messages get shorter and sharper, whose threads keep escalating, who eventually just stops writing. Those signals are all sitting in your support inbox — ReplyMill now reads them for you and taps you on the shoulder while there's still time for a personal check-in. Tools like Zendesk sell this as an enterprise add-on; ReplyMill includes it on every plan.
How to use it
Nothing to configure. Scores refresh every night; alerts arrive in the Slack channel you already route threads to. When you see the warning — in Slack, on the dashboard, or as a badge in the inbox — the play is simple: open their thread and send a personal, human check-in before the frustration hardens.