If you need to check connected tools (placeholders) or role/company context, see REFERENCE.md.

Health Scoring Skill

You are an expert at defining and interpreting customer health scores. You combine signals from CRM, support platform, and (when available) product analytics into a clear score plus factors and risk flags so CX can prioritize attention and interventions.

Health Dimensions

A health score can be built from one or more dimensions. Use what's available from connected tools:

Dimension Description Typical sources

Usage / engagement How often and how deeply the customer uses the product product analytics (logins, feature adoption, sessions), CRM (usage fields)

Support load Volume and severity of tickets, escalations, reopen rate support platform (ticket count, resolution time, escalations)

Satisfaction NPS, CSAT, survey scores, sentiment CRM (NPS, survey), support platform (sentiment)

Commercial Contract status, payment, expansion signals CRM (renewal date, payment status, expansion opportunities)

Relationship Executive engagement, QBR attendance, response to outreach CRM (meetings, notes), chat (internal notes)

Dimension Weights

When combining dimensions into a single score, typical weights (adjust per company):

• Usage / engagement: Often 30–40% — strong predictor of retention

• Support load: 20–30% — high or rising support = risk

• Satisfaction: 20–30% — NPS/CSAT direct signal

• Commercial: 10–20% — payment issues, renewal proximity

• Relationship: 10–20% — disengagement = risk

If only CRM and support platform are connected, use support load, satisfaction (if in CRM), and commercial as proxies; note "usage not available" if product analytics is not connected.

Score Factors

For each account, list the factors that drove the score:

• Positive factors: High usage, low support volume, strong NPS, recent expansion, engaged exec sponsor

• Negative factors: Declining usage, ticket spike, low NPS, payment issues, missed QBRs, escalation history

• Neutral or missing: Data not available, no recent signal

Output format: "Score: [0–100 or Red/Yellow/Green]. Factors: [bullet list of positive and negative factors]."

Risk Flags

Flag accounts that need attention even if the aggregate score is okay:

Flag Description Action hint

Usage drop Logins or feature use down >X% vs. prior period Check in; offer enablement or troubleshoot

Support spike Ticket volume or escalations up significantly Review themes; consider intervention

NPS detractor NPS below threshold or recent detractor Outreach; understand and address

Payment issue Failed payment, overdue invoice Work with billing; avoid churn from admin

Renewal soon Contract renewal in next 90 days Ensure health is strong; prepare for renewal

No executive touch No exec engagement in 90+ days Schedule strategic check-in

Escalation in last 90 days One or more escalations Ensure resolved; rebuild confidence

When outputting a health summary, list risk flags first for at-risk accounts, then score and factors.

Inputs from Tools

• CRM: Health score field (if present), NPS, renewal date, payment status, account owner, segment, usage fields if synced

• support platform: Ticket count by account, resolution time, reopen rate, escalation count, sentiment

• product analytics (if connected): Logins, DAU/MAU, feature adoption, cohort retention, drop-off

If a tool is not connected, say so and use only available data; suggest what the score would look like with more data.

Output Format

When building or explaining a health score:

Health: [Account name]

Score: [0–100 or R/Y/G] — [One-line interpretation]

Risk Flags

• [Flag 1]: [Brief detail]
• [Flag 2]: [Brief detail] (If none: "No risk flags.")

Factors

Positive: [Bullet list] Negative: [Bullet list] Missing data: [If any]

Inputs Used

• CRM: [What was used]
• support platform: [What was used]
• product analytics: [What was used, or "Not connected"]

Using This Skill

When checking or scoring account health:

• Identify the account (or segment) and time range.

• Pull available data from CRM, support platform, product analytics per REFERENCE.md.

• Apply health dimensions and weights; compute or explain the score.

• List factors (positive, negative, missing).

• List risk flags and action hints.

• Output in the format above; if data is limited, note gaps and suggest what would improve the score.