What it does
This skill runs your renewal playbook on a single account. You give it the renewal date and ARR, whatever evidence you have (a seat report, a support export, the last few call notes, the contact list with title changes) and the list of plays your team is allowed to run, and it returns a one-page brief: a Green, Amber, or Red risk read with the signals that drove it, a stakeholder status, a save plan of two to four sequenced plays with owners and dates, and the renewal ask in one sentence. It replaces the hour a CSM spends the week before a renewal opening five tabs and reconstructing the story from memory, and the outcome that hour usually produces, which is a vague sense that the account is fine.
The difference between a good version and a bad version is whether the brief argues or summarizes. A bad renewal brief lists everything it was given, calls the account 'at risk' without saying why, and proposes a discount because that is the play everyone remembers. A good one makes a call, names the two or three signals that forced it, tags each as Confirmed (present in the data) or Inferred (the model's reasoning), and picks plays that answer those specific signals in a sensible order: close the open escalation before you ask for an exec call, find the new owner before you run an adoption workshop for nobody. The playbook constraint is what keeps it honest. The skill cannot invent a play, so it cannot reach for a discount your team never approved.
It beats the manual version in three ways. Consistency: every renewal gets the same structure, so a CS lead can read twenty briefs on a Monday and compare them. Pattern-matching: the model connects a champion's title change in the contact list to a seat decline in the usage report and calls it an ownership problem, a link a busy CSM makes only after the renewal is lost. Sequencing: given a playbook, it orders plays by what has to be true first, and puts a date on each, so the brief is a plan and not a mood. For a Green account it flips to an expansion plan, because a healthy renewal with no ask is a missed quarter.
Where it does not help: it reasons only over what you paste. It has no view into your CRM, support desk, or call recorder on its own. That is the job of the paired use case, a weekly Claude agent that pulls the renewal window from those systems and hands each at-risk account to this brief. Run by hand, it is a two-minute prep tool for one account. Run inside that agent, it is the reasoning step that turns pulled data into a brief a CSM will act on. Either way, when data is missing it says Amber and names the gap, rather than filling the gap with a guess.
The renewal playbook template
The skill only proposes plays you have written down, so the playbook is the input that matters most. Copy the template below, fill in your risk definitions and the plays your team is actually allowed to run, and paste it into the Project or the SKILL.md folder alongside the skill. The numbered plays become the vocabulary of every brief.
RENEWAL PLAYBOOK, {{team}}
Owner: {{CS lead}} Last updated: {{date}}
RISK DEFINITIONS
- Green: seats active >= 70% of paid, no open escalation, champion in seat, exec touched this quarter
- Amber: any one of: usage down two months running, champion moved, open escalation, no exec touch in 6 months
- Red: two or more Amber signals, or a confirmed budget cut / competitor evaluation
PLAYS (what a brief may propose)
01 Adoption workshop: 60-min session on the two features the account pays for and does not use. Owner: CSM. Use when active seats < 60%.
02 Exec sponsor call: our VP to their economic buyer, agenda set by the brief. Owner: CS lead. Use when no exec touch in 6 months or champion moved.
03 New-champion onboarding: 1:1 with whoever inherited the account internally. Owner: CSM. Use when champion left or changed title.
04 Escalation close-out: engineering owner joins the call, written fix plan with dates. Owner: Support lead. Use when an escalation is open.
05 Multi-year offer: 2-year term at current price, price protection as the value. Owner: AE. Use only on Green accounts.
06 Scoped downsell: drop unused seats or a module, keep the logo, set an expansion trigger. Owner: AE. Use when budget cut is confirmed.
07 Discount: last resort, capped at {{x}}%, requires CS lead approval. Use only when price is the confirmed signal.
RULES
- A brief may propose plays 01-06 without approval. Play 07 needs a name on it.
- Every play has an owner and a date before the renewal.
- No play is proposed without the signal it answers.The rules at the bottom are the point. A playbook that lets a brief propose a discount first is a discount machine. Number the plays, give each a trigger and an owner, and put the discount last with a name on it.
Inputs & outputs
Inputs
- Account name, renewal date, and current ARR (required)
- Usage or adoption data: active vs paid seats, key-feature usage, the trend over the last quarter
- Support history: ticket count, open escalations, the tone of recent threads (an export or a summary)
- Notes or transcripts from the last 2-3 customer calls
- Stakeholder list with titles, and any title or company changes since the contract was signed
- Your renewal playbook: the numbered plays the brief is allowed to propose (template above)
Outputs
- A one-page brief: Risk read (Green/Amber/Red), Signals, Stakeholders, Save plan or Expansion plan, Renewal ask, What would change the call, Open questions
- Every signal tagged Confirmed (with its source) or Inferred, so the CSM knows what to verify on the next call
- A save plan of 2-4 plays taken only from your playbook, sequenced, each with an owner, a date, and the signal it answers
- A one-sentence renewal ask with the number, and the single condition that would change the risk call
- For Green accounts, an expansion or multi-year ask instead of a save plan
How to set it up
Decide where the skill lives
Two routes, both fine. In claude.ai, create a Project named something like "Renewals", paste the SKILL.md body below into its custom instructions, and add your filled-in renewal playbook as a Project file, so every chat in that Project already knows the plays. In Claude Code, save the SKILL.md in a skills/renewal-brief/ folder and keep the playbook next to it as playbook.md; the terminal example below passes both in. The Project route suits a CSM prepping accounts by hand. The folder route is what you want when the weekly renewal agent calls this skill for every at-risk account.
Write your renewal playbook once
Fill the template above before you run the skill on a real account. Define Green, Amber, and Red in numbers your team already tracks, then list the plays you are actually allowed to run, each with a trigger, an owner, and a place in the order. Put the discount last and give it an approver. This is a fifteen-minute exercise and it decides the quality of every brief afterward, because the skill will propose exactly these plays and no others. If your team has never written the plays down, the first draft of this document is worth more than the skill.
Run it with whatever evidence you have
Send: "Renewal brief for
Act on the first play the same day, then re-run before the call
The plan is sequenced for a reason: the first play is usually the one that makes the rest credible, like closing an open escalation before asking for an exec's time. Book it the day the brief lands. Then re-run the skill a week before the renewal conversation with the updated evidence, and watch the "what would change the call" line. If the number it named has moved, the risk color should move with it, and the renewal ask should change. If it doesn't, the playbook needs a sharper trigger, not the brief.
The SKILL.md
Save this as SKILL.md in a folder named renewal-brief, or paste the body into a Claude Project. Keep your filled-in renewal playbook alongside it; the skill proposes only the plays it finds there.
---
name: renewal-brief
description: Use when a CSM or AE preps one specific renewal. Takes the account's renewal date, ARR, usage, support history, call notes, stakeholders, and the team's renewal playbook; returns a one-page brief with a Green/Amber/Red risk read, a stakeholder status, a sequenced save plan drawn only from the playbook, and the renewal ask. Tags every signal Confirmed or Inferred; never invents usage, tickets, or quotes.
---
# Renewal Brief
You prepare one renewal. Output a one-page brief the account owner can read in two minutes and act on today. Turn scattered evidence into a risk call and a plan; do not summarize everything you were given.
## Inputs
- Account, renewal date, current ARR (required).
- Usage: active vs paid seats, key-feature use, quarter trend.
- Support: ticket count, open escalations, tone of recent threads.
- Notes or transcripts from the last 2-3 calls.
- Stakeholders with titles and any title or company changes.
- The renewal playbook: the numbered plays you may propose. If none is given, use a conservative default set and say so.
## Method
1. Tag each fact Confirmed (in the supplied data) or Inferred (your reasoning). Never invent a number, ticket, quote, or stakeholder move.
2. Make the call: Green (renews on current terms), Amber (at risk without action), Red (churn or heavy downsell likely). Name the 2-3 signals that drove it. Thin evidence means Amber plus what is missing.
3. Map stakeholders: champion status (in seat, moved, gone), economic buyer engagement, anyone new who could block or sponsor. A champion change is a top-line signal.
4. Build the save plan from the playbook only: 2-4 plays, sequenced, each with an owner, a date before the renewal, and the signal it answers. For Green, write an expansion or multi-year ask instead.
5. State the renewal ask in one sentence with the number.
6. End with the one thing that would change the call and 2-3 open questions.
## Hard rules
- No fabricated data. Missing is named as missing.
- Plays come from the playbook. No discount as the first play unless price is the confirmed signal.
- One page.
## Output template
**Renewal Brief: <Account>, renews <date>, ARR <amount>, [Green | Amber | Red]**
Sources: <inputs supplied>
**Risk read**
<2-3 sentences, signals tagged Confirmed/Inferred.>
**Signals**
- <signal> (Confirmed, <source>) / (Inferred)
**Stakeholders**
- Champion / Economic buyer / Watch: <name, title, status>
**Save plan** (or **Expansion plan**)
1. <play>, owner <name>, by <date>: answers <signal>
**Renewal ask**
<one sentence with the number.>
**What would change the call**
<one sentence.>
**Open questions**
- <question>
Run it from the terminal
Example run
A CSM preps a renewal ninety days out where the champion has moved roles and usage is sliding.
Prompt: renewal-brief. Account: Northwind Freight, renews 2026-11-30, ARR $84,000. Playbook: [pasted, plays 01-07]. Evidence: August seat report, Zendesk export Jun-Aug, transcripts from Jul 15, Aug 12, Aug 28 calls, contact list.
Output:
Renewal Brief: Northwind Freight, renews 2026-11-30, ARR $84,000, Red
Sources: seat report (Aug), Zendesk export (Jun-Aug), 3 call transcripts, contact list
Risk read
Two independent signals point the same way. Active seats fell from 62 to 41 of 80 paid between June and August (Confirmed, seat report), and the champion, Dana Ortiz, changed title from Director of Ops to VP Finance in July (Confirmed, contact list), so the person who bought this no longer owns the workflow. An escalation on route-import failures has been open 34 days (Confirmed, Zendesk). Nobody has asked for a price cut, so this is an adoption and ownership problem, not a price problem (Inferred).
Signals
- Active seats 62 -> 41 of 80, two months running (Confirmed, seat report)
- Champion moved to VP Finance in July; no replacement owner named (Confirmed, contact list)
- Escalation #4471 open 34 days, tone in the last thread is curt (Confirmed, Zendesk)
- Aug 12 call: ops manager mentioned "looking at what else is out there" (Confirmed, transcript)
Stakeholders
- Champion: Dana Ortiz, now VP Finance, out of the workflow, still signs the contract
- Economic buyer: Dana Ortiz (same person; a finance lens on an ops tool is a risk)
- Watch: Marcus Lee, Ops Manager, ran the last two calls, not yet a champion
Save plan
1. Escalation close-out (play 04), owner Support lead, by 2026-09-19: answers the 34-day escalation before anything else is credible
2. New-champion onboarding with Marcus Lee (play 03), owner CSM, by 2026-09-26: answers the ownership gap
3. Adoption workshop on route import and dispatch board (play 01), owner CSM, by 2026-10-10: answers the seat decline
4. Exec sponsor call to Dana Ortiz, framed on cost per active seat (play 02), owner CS lead, by 2026-10-24: answers the finance lens
Renewal ask
Renew flat at $84,000 with a seat true-up clause at month six, once active seats are back above 55.
What would change the call
Active seats back above 55 by mid-October moves this to Amber.
Open questions
- Who inside ops now owns the dispatch workflow day to day?
- Is the "what else is out there" comment a formal evaluation or frustration with the escalation?
- Does Dana still see this line item as hers to defend in the FY27 budget?Related plays
- Churn Risk, Scored on Evidence: A Weekly Claude Agent for Renewals 90 Days Out →The paired play: an agent that pulls the whole renewal window every Monday and hands each at-risk account to this brief.
- QBR Prep on Autopilot: A Claude Agent Fills the Template from Your CRM, Calls, and Tickets →The same evidence, assembled into the quarterly conversation this brief protects.
- Claude + HubSpot integration guide →Where the renewal dates, contacts, and activity log come from when you automate the pull.
- Claude Projects: persistent, context-rich workspaces →How to keep the playbook loaded so every brief already knows the plays.