At-Risk Renewal Save Desk

Inside this Project, Claude carries the retention picture:

• The risk-scoring model — the signals, weights, and thresholds, so tiers mean the same thing across accounts and across quarters.
• Each account's risk history and trajectory — a tier-three account worsening is a different situation from one improving.
• Which save plays were run, for which accounts, and what happened.
• What has actually worked, which is the only reliable guide to what to try next.

Run one Project for your whole book. The comparative history is the value; one Project per account would throw it away.

Reach for this Project when:

• An account is showing risk signals ahead of renewal.
• You need to triage a set of renewals and decide where the effort goes.
• A champion has left and you need a re-engagement approach.
• An account needs executive escalation and you need a brief that earns the attention.

Use something else when:

• The account is healthy and the conversation is a business review — that is QBR Builder for Customers.
• You are managing pipeline rather than a book of renewals — that is Pipeline Review Workspace.

Per account, you provide:

• The account name, renewal date, and contract value.
• Usage trend over at least two comparable periods — a single snapshot cannot show direction.
• Support and sentiment signals: ticket volume, severity, escalations, survey scores.
• Champion status — still present, changed role, or departed.
• Anything the customer has said about renewal, budget, or their own priorities.

Two comparable periods is the minimum. Without a trend, seasonal variation is indistinguishable from decline, and that mistake produces false alarms that erode trust in the whole model.

The save-play debriefs are what make this Project improve. Most retention programmes never write them, which is why the same play gets run repeatedly without anyone knowing whether it works.

Claude accepts files up to 30MB each, and there is no fixed limit on how many you add, but everything in a Project's knowledge has to fit the context window. Fewer, sharper files beat a full archive.

Prepare files this way:

• Strip cover pages, legal boilerplate, and navigation chrome — they consume context and carry no signal.
• Prefer text formats. A clean export beats a scanned PDF.
• Name each file for what it contains, not when you exported it. Claude reads filenames.
• One subject per file. A merged reference document is harder to cite precisely.
• Date anything that goes stale, inside the file itself, so Claude can tell you when it is working from old material.

1. Create the Project. Name it for your book — for example Save Desk — Enterprise West.
2. Paste the Project Instructions.
3. Replace the three bracketed fields. The health-score definition matters most: if your organisation already has one, the rubric should defer to it rather than compete with it.
4. Upload the knowledge files, escalation thresholds and renewal calendar first.
5. Triage one account you already understand well. If the rubric disagrees with your read, that is the calibration conversation — adjust the weights in your health-score placeholder rather than ignoring the output.
6. Write the debrief after every save attempt, won or lost. This is the step that compounds.

The bracketed fields below are placeholders you replace with your own details before you save the instructions. Leaving them unfilled produces generic output — that is the single most common reason a Project underperforms in its first week.

This Project has three:

• [FILL IN: Company Name] — your employer.
• [FILL IN: Your Health Score Definition] — how your organisation already measures account health, if it does. The rubric below defers to yours where they overlap; leaving this unfilled means the default weights apply and Claude will say so.
• [FILL IN: Your Escalation Thresholds] — the contract value, risk tier, or circumstance that triggers executive involvement, and who gets involved. Without it, escalation recommendations are generic and get ignored.

Role

You are a retention analyst for [FILL IN: Company Name], running a save desk across a book of renewals. You triage risk on evidence, design save plays with named owners, and never propose a commercial concession.

Operating rules

• Never assign a risk tier without showing the per-signal scoring.
• Always distinguish usage decline from seasonal variation.
• Never draft a customer-facing concession — route it to the account owner.
• Never score a signal with no data; report it unscored and reduce the total.
• Never present a save play as a prediction of retention.

The risk rubric — five weighted signals

Usage trend (30) · champion status (25) · support and sentiment (20) · business-outcome delivery (15) · commercial signals (10). Tier 1 critical at 80+, tier 2 high at 60–79, tier 3 watch at 40–59, tier 4 stable below 40.

The full instructions continue with the scoring guidance per signal, the four output templates, the commands, and escalation.

Four modules, each with a defined output template:

• Risk triage — a tier with the per-signal scoring shown.
• Save play — actions with owners and dates.
• Re-engagement sequence — for a lapsed or changed relationship.
• Executive escalation brief — short enough that an executive reads it.

A risk triage summary. Calderwood Insurance is fictional.

Risk triage — Calderwood Insurance · renewal 12 November · $186k · 94 days out

Signal scoring

Three contributing signals, in order

1. Champion departure with no re-established relationship. Ten weeks since departure, no executive relationship with the successor. In the debrief history, champion loss without re-engagement inside 60 days is the strongest predictor of non-renewal in this book.
2. Sustained usage decline that is not seasonal. 34% over two quarters against a flat comparable period last year. The decline concentrates in one business unit, which suggests a specific cause rather than general disengagement — worth finding.
3. Procurement pricing request. Ambiguous alone; meaningful alongside the other two.

Save play

Escalation trigger

Per the thresholds, tier 2 above $150k with fewer than 100 days to renewal warrants executive sponsorship now rather than at tier 1. Recommend engaging the executive sponsor this week.

No commercial concession is proposed here. If the successor champion raises price, that is a conversation for the account owner and finance, not a save play.