Who actually uses it, who actually pays for it, who left

What is the exact prompt?

Run it 90 days before the renewal date, not the week before:

• "Build an account map for [company], renewal on [date]. From the CRM: every contact on the account with title, role in the deal if recorded, and the date of our last touch with each. From email: search my history with their domain and list everyone who has emailed with us in the last 12 months who is NOT in the CRM, and everyone in the CRM who has gone silent for 90+ days. Flag any contact whose emails started bouncing. That usually means they left."
• "From product analytics: monthly active users on their account for the last 6 months, trend direction, which teams or workspaces are active, and which features they use most. Note anything that dropped sharply and when."
• "Output one page: a stakeholder table (name, title, role, last touch, status: engaged / silent / departed / unknown), a usage paragraph with the trend, three risks ranked with the evidence for each, and the two people we should talk to this month and why. Cite a source for every claim. Read-only everywhere: change no records."

Cross-reference email against the CRM

The CRM records who you were supposed to know.

Email records who you actually talked to. The delta is the valuable part of the map: the power user who emails support weekly and was never added as a contact, and the economic buyer whose last reply was in another quarter.

Bounce detection pays first. A bounced champion is a five-alarm renewal risk, knowable months before a job title changes. And a stakeholder table sorted by last touch is a to-do list for the quarter.

What do the risks look like when the evidence is real?

Like this: "champion risk: Dana's last reply was March 14, her emails now bounce, and the workspace her team used is down 60% since April". Requiring evidence per risk is also what stops the generic renewal-risk listicle that applies to every account and helps with none.

Usage does the same work on the health side. A flat headline number hides one team quintupling while two others go dark, so the by-team breakdown from Amplitude or PostHog tells you whether this is an expansion conversation or a rescue, which is the daily bread of customer success teams.

Two names to call, then a monthly diff

The map ends with two names on purpose.

A one-page brief that says "talk to these two people this month, for these reasons" gets acted on.

Re-run it monthly and diff against the last run: status changes and usage inflections are the early warning, the same rhythm as churn analysis at one account. Record fixes stay a separate batched job, covered in the CRM data-entry playbook.

Frequently asked questions