Customer Health Scores: The Early Warning System Your Business Needs
By the time churn shows up in your revenue numbers, it is too late. Customer health scores give you weeks or months of advance warning. Here is how to build one that actually works.
Every SaaS company has a moment where they stare at their churn numbers and ask: "How did we not see this coming?"
The answer is almost always the same: they were looking at the wrong metrics, or looking at the right metrics too late. Monthly churn rate is a rearview mirror. It tells you what already happened. Customer health scores are a windshield. They tell you what is about to happen.
What a Health Score Actually Is
A customer health score is a composite metric that combines multiple signals into a single number (typically 0-100) representing how likely a customer is to remain active and grow over time.
Think of it like a credit score for customer relationships. A credit score does not measure one thing. It combines payment history, credit utilization, account age, and other factors into a single number that predicts future behavior. A health score does the same for customer retention.
The Inputs That Matter
Not all signals are equally predictive. Based on analysis across hundreds of SaaS products, these are the inputs that consistently correlate with retention:
Usage Signals (Weight: 40%)
Login frequency relative to baseline. Not absolute logins, but logins compared to the user's own history. A user who typically logs in daily and drops to twice a week is at higher risk than a user who has always logged in twice a week.
Feature breadth. How many distinct features does the user engage with? Users who use 4+ features churn at 3x lower rates than users who use only 1-2.
Core action completion. Every product has 2-3 actions that represent genuine value delivery. For a messaging platform, it might be "sent a campaign" or "created an automation." Track whether these core actions are happening at a healthy frequency.
Engagement Signals (Weight: 25%)
Message engagement. Are they opening your emails? Clicking your push notifications? Engaging with in-app messages? Declining engagement with your communications often precedes declining engagement with your product.
Support interactions. Support tickets are not inherently negative. In fact, users who submit support tickets and receive good responses are often more loyal. What matters is the trend: increasing ticket volume with declining sentiment is a red flag.
Community participation. If you have a community (Slack, Discord, forum), participation is a strong positive signal. Users who are part of a community churn at 40% lower rates.
Relationship Signals (Weight: 20%)
Team adoption. How many team members are actively using the product? Single-user accounts are far more vulnerable than multi-user accounts.
Integration depth. How many integrations has the user set up? Each integration increases switching costs and makes the product more embedded in their workflow.
Account age. Older accounts are generally more stable, but only if usage remains healthy. An old account with declining usage is a different kind of risk than a new account with low usage.
Financial Signals (Weight: 15%)
Plan tier. Higher-plan customers have made a larger financial commitment, which correlates with lower churn intent (though not always lower actual churn if value is not delivered).
Payment history. Failed payments, downgrade requests, or delayed renewals are strong churn indicators.
Expansion history. Has the customer upgraded, added seats, or purchased add-ons over time? Expansion is the strongest signal of a healthy relationship.
Building the Score
Step 1: Normalize Each Input
Each input should be normalized to a 0-100 scale based on your own data. "Good" and "bad" look different for every product.
For login frequency: if your median user logs in 15 times per month, a user logging in 20+ times scores 100, while a user logging in 3 times scores 20.
Step 2: Weight by Predictive Power
Not all inputs are equally predictive of churn. Run a correlation analysis against your historical churn data to determine weights. The weights we listed above are starting points, but your actual weights should be calibrated to your own data.
Step 3: Calculate and Categorize
Combine the weighted scores into a single number. Then categorize:
- 80-100: Healthy. Low churn risk. Focus on expansion opportunities.
- 60-79: Stable. Moderate risk. Monitor for declining trends.
- 40-59: At Risk. Elevated churn probability. Proactive intervention needed.
- 0-39: Critical. High churn probability. Immediate human outreach required.
Step 4: Automate Responses
Each category should trigger a different engagement strategy:
Healthy (80-100): Quarterly check-in. Expansion suggestions. Referral asks. Let them be; they are happy.
Stable (60-79): Monthly feature education. Personalized tips. Gentle deepening of engagement. The goal is to move them toward Healthy, not just maintain.
At Risk (40-59): Weekly automated touchpoints. Re-engagement sequences. Surface value they may have forgotten. Offer to schedule a call with customer success.
Critical (0-39): Immediate human outreach. A personal email or call from customer success. Understand what changed. Offer concrete solutions to specific problems. This is triage.
Common Mistakes
Mistake 1: Weighting by Gut Instead of Data
Many teams set health score weights based on intuition: "I think usage is the most important factor." Run the actual correlation analysis. You might discover that support sentiment is a stronger predictor than login frequency for your specific product.
Mistake 2: Not Updating the Model
Customer behavior patterns change over time. A health score model calibrated on 2024 data may not be accurate in 2026. Recalibrate quarterly by testing your model's predictions against actual churn.
Mistake 3: Treating the Score as a Number Instead of a Trend
A score of 65 is concerning if it was 85 last month. A score of 65 is encouraging if it was 45 last month. Always display the score alongside its recent trend. A declining healthy account may need more attention than a stable at-risk account.
Mistake 4: Ignoring the Score
This is the most common mistake. Teams build health scores, display them on a dashboard, and then do nothing with them. A health score without automated responses and human workflows is just a number. It needs to trigger action.
The ROI of Health Scoring
Companies that implement health scoring and respond to the signals typically see:
- 25-40% reduction in churn among accounts flagged as At Risk or Critical
- 15-20% increase in expansion revenue from proactive outreach to Healthy accounts
- 50% reduction in surprise churn (churn from accounts that were not identified as at-risk)
The last number is the most valuable. "Surprise churn" is churn that nobody saw coming, and it usually means lost revenue, a scramble to understand what went wrong, and a customer who will never come back.
A health score will not catch every at-risk customer. But it will catch most of them, early enough to do something about it. And in the retention game, early warning is everything.