Customer Health Score
Also known as: Customer Health Index, Account Health
A composite score that evaluates the overall health of a customer relationship, typically combining product usage, engagement quality, support sentiment, and business outcomes into a single indicator.
Formula
Varies — typically a weighted combination of: Product usage signals (30-40%) + Engagement frequency and quality (20-30%) + Support ticket sentiment and volume (15-20%) + Business outcome achievement (15-20%) Who Is This Metric For?
Check daily to identify at-risk accounts and prioritize outreach to declining health scores.
Review weekly to spot portfolio-level health trends and allocate resources to struggling segments.
Track aggregate health distribution as a leading indicator of future retention and expansion.
Correlate health score components with product usage to identify features that drive customer success.
Priority by Stage
Start simple. Even a 3-color RAG status (red/amber/green) based on 3 signals is better than nothing. Don't over-engineer it.
Build a real health score with 5–7 weighted inputs. The key is consistency — every CSM should assess health the same way.
Health scores should be automated and trigger workflows. Focus on predictive accuracy — does your score actually predict churn?
Health scoring should be ML-enhanced, validated against outcomes, and continuously calibrated.
Health Score Components by Stage
Crawl
Recommended Inputs
- Product login frequency
- Support ticket volume
- CSM sentiment (gut feel, documented)
Scoring Method
Simple RAG (Red/Amber/Green) updated monthly by CSM
What to Skip
Don't try to build a 10-factor weighted model. It will be inaccurate and nobody will maintain it.
Walk
Recommended Inputs
- Product usage depth (feature adoption, not just logins)
- Engagement recency and quality
- Support ticket sentiment
- Stakeholder relationship strength
- Contract and renewal status
Scoring Method
Weighted score (0–100) with defined thresholds for healthy/at-risk/critical
What to Skip
Don't add NPS as a health input yet unless you have reliable, frequent NPS data. Annual NPS is too lagging.
Run
Recommended Inputs
- Product usage trends (not just current state)
- Multi-stakeholder engagement map
- Support trend analysis
- Business outcome achievement vs. success plan
- Expansion readiness signals
- Executive sponsor engagement
Scoring Method
Automated composite score with trend indicators and automated alerts at threshold crossings
What to Skip
Don't rely solely on automated scores — keep a CSM override mechanism for qualitative signals the data can't capture.
Fly
Recommended Inputs
- All Run-stage inputs plus: Predictive churn probability
- Customer effort score trends
- Product usage relative to peer cohort
- Advocacy and referral activity
- Multi-product adoption depth
Scoring Method
ML-driven predictive health with cohort benchmarking, anomaly detection, and confidence intervals
What to Skip
Don't black-box the model. CSMs need to understand why the score is what it is, or they won't trust it.
Common Mistakes
- Building a complex health score before you have reliable data inputs — garbage in, garbage out
- Making health scores entirely subjective (CSM opinion) or entirely objective (data only) — the best scores blend both
- Not validating the score against actual outcomes — if red accounts aren't churning more than green accounts, your score is broken
- Updating health scores too infrequently — monthly is minimum, weekly is better for high-touch accounts