| name | churn-save-plan |
| description | Deep-dives a red account and produces a structured save plan — root causes ranked, save plays matched to causes, and a realistic assessment of whether the account is savable. Use when an account turns red, gives churn signals, or someone says "we're going to lose X". |
| license | MIT |
| compatibility | Tool calls resolve natively in Quivly, or in Claude Code / claude.ai / any MCP-enabled agent via the Quivly MCP connector. Without tools, the skill degrades to a guided manual workflow. |
| metadata | {"author":"quivly","version":"1.0","category":"customer-engineering","quivly-tools":"health.get-health-score health.get-insights platform-data.search-tickets communication.search-calls platform-data.get-usage customer-data.get-contacts"} |
Churn Save Plan
You build the save plan for an account in real danger — honest about root causes, honest about odds, specific about plays.
Core principle: no play without a named root cause — discounts don't fix value gaps, and generic plays waste the runway.
Workflow
- Establish the full risk picture: health components (
get-health-score), active signals (get-insights), ticket history (search-tickets), usage trend (get-usage), recent call tone (search-calls), relationship map (get-contacts).
- Identify root causes — usually 1-2 dominate. Rank them:
- Value gap — never reached the outcome they bought for
- Product gap — a needed capability doesn't exist or underperforms
- Trust damage — support failures, missed commitments, outage scars
- Relationship loss — champion gone, sponsor disengaged, single thread cut
- External — budget cuts, M&A, strategy shift (least addressable by us)
- Assess savability honestly: is there time before the notice deadline, an engaged human on their side, and a cause we can actually fix?
- Build the plan: each play matched to a root cause, with owner, deadline, and a "proof point" — the observable event that tells us it's working.
Output Format
Save plan: {Customer} — ARR, renewal/notice date, savability: Good odds / Uphill / Long shot
Root causes — ranked, each with evidence
The plan — table: play, targets which cause, owner, deadline, proof point
Ask list — internal help needed: exec sponsor, product commitment, commercial flexibility
Tripwire — the date/event at which we stop investing in the save and shift to graceful-exit + win-back planning
Stop conditions
- Two re-engagement attempts with no response → stop sending. The next move is an exec-to-exec touch or the tripwire decision, not a third email.
- The tripwire date or event passes → stop the save. Shift to graceful exit and win-back planning.
- No engaged customer contact AND no exec path in → the plan is not viable as written; say so and escalate the relationship question before proposing plays.
Self-check
- The plan has five plays → you don't know the root cause yet. Go back to diagnosis.
- A discount appears before a named cause → you're buying time, not saving the account.
Guidelines
- One play per root cause beats five generic plays. Discounts don't fix value gaps.
- A save plan without an engaged customer contact is a fantasy — if no one on their side will meet, that IS the first play.
- Set the tripwire on day one. Zombie save plans consume the team's best hours.
- "Long shot" is a legitimate verdict; say it and size the effort accordingly.
Related skills: diagnosis inputs → health-drop-diagnosis, usage-drop-investigation, ticket-pulse; exec entering the account → escalation-brief.