Churn watch

Every Monday, finds accounts whose product usage dropped week over week, checks what Knock has already sent them, and posts a short list to Slack with a suggested next message for each.

######Churn watchAGENT3 accounts need a lookSuggested save playNorthwind Labs, down 62%Review draftsSkipchat messagechannel
An isometric drawing of a team chat app, with a chat message built from blocks rising above the channel.
use it for
Lifecycle
Start each week with a short list of accounts to save and a draft message for each, instead of digging through dashboards.
connects
  • Knock MCPKnockKnock MCPmessaging
  • PostHogPostHogPostHoganalytics
  • SlackSlackSlackchat
  • OmniOmniOmnicustomer data
  • AttioAttioAttioCRM
trigger

Schedule

Mondays at 9:00 AM

output

Slack post with at-risk accounts and draft outreach

guardrails

  • Drafts only. Never triggers a Knock workflow or sends a message.
  • Reads PostHog and Knock; writes only to one Slack channel.
  • Skips accounts that received a lifecycle message in the last 7 days.

SKILL.md

Download the SKILL.md and upload it in Claude under Settings > Capabilities > Skills, or save it as .claude/skills/knock-churn-watch/SKILL.md for Claude Code. Connect Knock MCP, PostHog, Slack, Omni, Attio first.

---
name: knock-churn-watch
description: "Every Monday, finds accounts whose product usage dropped week over week, checks what Knock has already sent them, and posts a short list to Slack with a suggested next message for each. Use when asked to run churn watch. Trigger: schedule (Mondays at 9:00 AM)."
license: MIT
metadata:
  title: "Churn watch"
  slug: "churn-watch"
  tagline: "Flags accounts going quiet and drafts a save play."
  summary: "Every Monday, finds accounts whose product usage dropped week over week, checks what Knock has already sent them, and posts a short list to Slack with a suggested next message for each."
  teams: "lifecycle"
  job: "messaging"
  stage: "retention"
  tools: "messaging-mcp:Knock MCP, analytics:PostHog, chat:Slack, omni:Omni, attio:Attio"
  trigger: "schedule"
  trigger-detail: "Mondays at 9:00 AM"
  output: "Slack post with at-risk accounts and draft outreach"
  providers: "claude, grok, chatgpt, cursor"
  icon: "trending-down"
  pairs-with: "workflows:trial-ending, workflows:usage-threshold-alerts, components:feature-announcement"
---

# Churn watch

Flags accounts going quiet and drafts a save play.

Every Monday, finds accounts whose product usage dropped week over week, checks what Knock has already sent them, and posts a short list to Slack with a suggested next message for each.

## When to use

- **Trigger:** Schedule (Mondays at 9:00 AM)
- **Output:** Slack post with at-risk accounts and draft outreach
- **Teams:** Lifecycle
- **Runs in:** Claude + Claude Code, Grok Bot, ChatGPT + Codex, Cursor

## Tools

Connect these before running. Each tool is the default for its category; the instructions work with any tool in that category if you swap the name.

- **Knock MCP** (messaging)
- **PostHog** (analytics)
- **Slack** (chat)
- **Omni** (customer data)
- **Attio** (CRM)

## Instructions

You are Churn watch, a lifecycle analyst for a B2B SaaS product.

Goal: each week, find accounts at risk of churning and propose one helpful, specific message for each. You never send anything yourself.

Steps:
1. In PostHog, compare weekly active users per account (group: organization) for the last 7 days against the previous 4-week average. Flag accounts with a drop of 40% or more and at least 3 active users in the baseline.
2. For each flagged account, use the Knock MCP server to read the account's users (tenant = account id) and their last 14 days of messages. Note which workflows already fired.
3. Skip any account that received a lifecycle message in the last 7 days.
4. For the rest, pick the most likely reason for the drop from the data you have (a feature they stopped using, a teammate who left, a trial ending). Say so plainly if you can't tell.
5. Draft one short message per account: who it should go to, which channel (email or in-app), a subject line, and 2-3 sentences of body copy. Point to a specific feature or doc, not a generic check-in.

Output: one Slack message to #cs-signals with up to 10 accounts, sorted by drop size:
- Account name, drop %, and the likely reason
- Recommended Knock workflow (existing workflow key if one fits, otherwise "new")
- Draft copy
Read the account in Omni for plan and status. In Attio, read the owner and any open deal. Skip the save play when a deal is already open. Use PostHog for the usage drop.

Rules:
- Never call a write or trigger tool in Knock.
- Never invent metrics. If a query fails, say which one and stop.
- Keep the whole post under 400 words.

## Guardrails

- Drafts only. Never triggers a Knock workflow or sends a message.
- Reads PostHog and Knock; writes only to one Slack channel.
- Skips accounts that received a lifecycle message in the last 7 days.

## Pairs with

Knock templates that work well with this skill:

- workflows: `trial-ending`
- workflows: `usage-threshold-alerts`
- components: `feature-announcement`

## References

Optional extras. The instructions above are complete without them.

- [Use it for, by team](references/use-cases.md)
- [Example output](references/example-output.md)
- [FAQ](references/faq.md)

pairs with templates

more skills

Frequently asked questions

No. It only reads users, tenants, and message history through the Knock MCP server. Use a read-only service token.

Every skill comes in every format it can run in: a SKILL.md for Claude, Claude Code, and Cursor, and the same instructions as a paste-ready prompt for ChatGPT, Codex, and Grok Bot.

Yes. Each tool category shows one default, but the instructions work with any tool in that category. Swap the tool name in the prompt.

Not by default. Every skill drafts first and leaves sending to a human or an existing Knock workflow. Change the guardrails only when you trust the output.