Customer Health Scores Are Broken

Every CS platform sells you the same pitch: aggregate product usage data into a "health score," color-code it green/yellow/red, and you'll predict churn. It sounds logical. Customers who don't log in are at risk. Customers who use every feature are healthy.

Except it doesn't work like that in service businesses.

If you sell a service (not software), your customers don't generate login events or feature adoption metrics. They generate support emails, phone calls, and service requests. The traditional health score model wasn't built for you, and forcing it to fit creates blind spots that cost accounts.

The Three Blind Spots of Usage-Based Health Scores

1. Active users can be furious

A customer who logs into your portal daily to check status updates might look "healthy" by every usage metric. But if they're logging in because their issue hasn't been resolved in two weeks, they're one bad interaction away from filing a complaint.

High activity doesn't mean high satisfaction. Sometimes it means high frustration.

The usage score says "engaged." The communication analysis says "preparing to escalate." One of these is right. The other one gets you a BBB complaint.

2. Silent customers aren't safe

The opposite is also true. A customer who never contacts support might be perfectly satisfied. Or they might have given up. Usage scores can't distinguish between "happy silence" and "pre-departure silence."

Communication analysis can. A customer who went from 3 emails/month to zero after an unresolved issue has a very different risk profile than one who's never needed help.

3. Sentiment isn't a feature you can track

Product analytics can tell you what a customer did. They can't tell you how the customer feels. And in service businesses, feelings drive decisions. A customer who feels ignored will leave. A customer who feels heard will stay through imperfect service.

The only way to measure sentiment at scale is to analyze what customers actually say. Not their clicks. Their words.

What Service Businesses Actually Need

Usage signals work for SaaS companies where the product is the entire relationship. For service businesses, the relationship lives in conversations. That's where the signals are.

The Communication-First Health Model

Here's what a health score should look like for a service business:

1. Analyze every support interaction. Emails, call transcripts, chat messages, task notes. Not just CSAT surveys after the fact.
2. Score sentiment and intent in real time. Not "positive/negative" binary. Granular: frustrated, threatening, documenting, escalating, disengaging.
3. Track trajectory, not snapshots. A customer at risk score 40 who was at 20 last week is more concerning than a customer sitting steady at 60. Direction matters more than position.
4. Flag specific signals, not just numbers. "Score dropped to 45" is useless to a CSM. "Customer referenced BBB in their last two emails and CC'd their attorney" is actionable.
5. Deliver alerts when they matter. Not in a weekly report. Within minutes of a high-risk interaction. The intervention window is hours, not days.

Why This Matters for SMBs

Enterprise CS teams have Gainsight, ChurnZero, and Totango. These tools cost $30K-$150K/year and are built for SaaS companies with millions of usage events.

SMB service businesses (50-200 employees) have a different problem:

• They don't generate product usage data (or very little)
• Their customer relationships are conversation-heavy
• They can't afford enterprise CS platforms
• They need something that works with what they already have: an inbox

The gap in the market isn't "cheaper Gainsight." It's a fundamentally different approach that starts with communication analysis instead of product analytics.

What This Looks Like in Practice

Imagine your support team handles 200 emails a day. In a traditional setup, tickets get prioritized by age, SLA, or manual triage. The angry email from your highest-value account gets the same queue position as a routine question.

With communication-based risk scoring:

• Every email gets analyzed as it arrives
• High-risk communications (score 70+) get flagged immediately
• Your CS director gets a Slack alert: "Account #4521 - risk score jumped from 30 to 82. Last email contains escalation language and BBB reference."
• Within an hour, a senior team member is on the phone
• The complaint that would have been filed Thursday never happens

That's not a health score. That's an early warning system. And for service businesses, it's what "customer health" should have meant all along.

Key Takeaways

• Usage-based health scores are built for SaaS, not service businesses
• Active users can be furious. Silent customers can be departing. Usage data can't tell the difference.
• Service businesses generate their most valuable signals in support conversations, not product analytics
• Communication-based risk scoring analyzes what customers say, not what they click
• The SMB market (50-200 employees) is underserved. Enterprise tools are too expensive and designed for the wrong data.