Churn risk review

Churn risk review

A weekly 30-minute review that ends with every at-risk account having one owner, one play, and one date. The point is decisions, not a longer list.

Inputs

Use whatever exists; ask for the rest. In order of value:

1. At-risk list: out/health.csv from the customer-health-score kit, or a list of accounts the team is worried about.
2. Usage: out/usage-dashboard.json (accounts trending down) from saas-usage-dashboard, or any usage export.
3. Renewal and money: ARR, renewal date, auto-renew, payment status (days past due).
4. Support: open and urgent tickets, recent CSAT.
5. Notes: CRM notes, last call summary, champion changes, any "we are evaluating alternatives" signals. Ask the account owner for one line each if nothing is written down.

Keep the review to the accounts that matter: red accounts, yellow accounts that dropped 10+ points, and anything renewing in the next 90 days with a warning sign. Cap it at 10 accounts; if there are more, sort by ARR at risk and say what was left out.

Steps

1. Build the list. One row per account: name, ARR, renewal date, score and change, and the evidence (numbers, not adjectives).

2. Name the risk driver for each account. Pick the one that, if fixed, would most likely keep them. Use this set:

   Driver	Typical evidence
   No value yet	low seat use from the start, core feature unused, never activated
   Value fading	usage trend down, fewer active users, key workflow stopped
   Champion lost	main contact left or changed role, no exec sponsor
   Product gap	tickets or notes about a missing feature, competitor mentioned
   Support pain	urgent tickets, low CSAT, slow responses
   Budget or price	downgrade asks, procurement review, past due invoices
   Business change	layoffs, acquisition, project cancelled

   If evidence is thin, say "Unknown, needs a call" instead of guessing.

3. Pick a save play that matches the driver:

   • No value yet: a working session to set up the one workflow they bought it for; success criteria in writing.
   • Value fading: usage review with the admin; show what dropped and offer training for new team members.
   • Champion lost: find and meet the new owner within two weeks; exec-to-exec note.
   • Product gap: honest roadmap answer, a workaround, or a product call; log the gap for the product team.
   • Support pain: escalate the open issues, name one support contact, follow up within 48 hours.
   • Budget or price: right-size seats or plan before they ask to cancel; trade term length for price only with approval.
   • Business change: confirm the new situation early; offer a smaller plan rather than losing the logo.
4. Assign an owner and a due date (this week or next) for each play, plus the signal that would show it worked.
5. Write out/churn-risk-review.md with the template below, and out/churn-risk-review-agenda.md for the meeting.
6. Next week, open last week's file first and mark each play done, slipped, or no longer needed before adding new accounts.

Review template

# Churn risk review: week of <date>

ARR at risk: $X across N accounts (M renewing in 90 days)

| Account | ARR | Renews | Score | Driver | Evidence | Play | Owner | Due | Worked if |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Bluebird Labs | $42,000 | Dec 15 | 25 (-39) | Value fading | events -68% in 28 days, 35% seat use, 2 urgent tickets | Usage review with admin, retrain new team | CSM | Nov 8 | weekly active users back to 5+ |

## Decisions needed
- <anything that needs a manager, discount approval, or product commitment>

## Last week's plays
- <account>: done / slipped / dropped, result

Agenda template (30 minutes)

1. Last week's plays: 5 min, status only
2. New and worsening accounts: 15 min, biggest ARR first, agree driver and play
3. Decisions and escalations: 5 min
4. Patterns: 5 min (same driver in 3+ accounts? tell product or support leads)

Guardrails

• Never contact a customer, change a subscription, or offer a discount. This kit writes a plan; people act on it.
• Evidence must be facts from the data or a named person's note. Mark guesses clearly as guesses.
• Keep customer contact names and ticket text out of chat; reference accounts by name only.
• A discount is never the first play. If the user wants to lead with one, note it under "Decisions needed" for approval.

Churn risk review

Runs your weekly churn risk review. Claude pulls together health scores, usage, renewals, support, and notes for the accounts that matter, names the real risk driver for each, and writes a one-page review where every account has a save play, an owner, and a date.

Who it is for: customer success leads, account managers, and founders who own renewals.

Example

Input: out/health.csv from the customer-health-score kit, plus a line of notes per account from the account owner.

Output: out/churn-risk-review.md (see examples/review-example.md):

And out/churn-risk-review-agenda.md for a 30-minute meeting.

What it can access

• Instructions only, no scripts. Works in Claude Code, Codex, and Cursor.
• Reads data/ and out/ (exports and other kits' outputs). Writes out/churn-risk-review*.
• No network. It never contacts customers or changes anything in your tools.

Setup

None. Pairs well with @usagemetrics/customer-health-score and @usagemetrics/saas-usage-dashboard.