Every SaaS company has a health score somewhere. Usually it's a column in the CRM. Green, amber or red.
And every customer success team has a story about the account that was green on Monday and gone by Friday.
The problem isn't the idea. Knowing which customers are healthy is exactly right. The problem is that most health scores measure the wrong things, update too slowly and get ignored by the people they're meant to help.
Why health scores miss
They measure activity, not value
Logins, page views and email opens are easy to count. They're also easy to misread.
A customer who logs in every day might be stuck, clicking around trying to make something work. A customer who logs in twice a month might be perfectly happy, because that's all the job needs. Payroll software doesn't need daily visits to be loved.
Customer success platform ChurnZero makes this point well: activity can look busy while delivering very little, and quiet accounts can be thriving.
They trust one person too much
One enthusiastic power user can make a whole account look healthy. Lots of activity, great engagement, solid green.
Then that person leaves. Suddenly nobody else knows how to use the product, and the account goes from green to gone. Investor David Skok names the loss of the champion as one of the two biggest causes of SaaS churn. A score that counts total activity can't see it coming.
They ignore the relationship
Renewals are decided by people. Is the executive sponsor still engaged? Does the champion reply to emails? Are there two or three contacts, or just one?
Those signals often move before usage does, and most scores don't track them at all.
They go stale
Plenty of health scores were designed in a workshop two years ago, weighted by gut feel and never checked against which customers actually churned. A score nobody validates is a guess with a colour on it.
The signals that matter
A useful health view combines four kinds of signal, tracked over time rather than as a snapshot.
1. Value usage. Are they using the features tied to why they bought? If they bought for reporting, are reports being built and shared? Depth beats volume.
2. Breadth. How many people use the product, and is that number growing or shrinking? An account with one active user is one resignation away from churn.
3. Payments. Failed cards, downgrades, seats removed, a switch from annual to monthly. Money signals are honest signals.
4. Relationship. Is the champion still there and still responsive? Is anyone senior engaged? When did a person last have a real conversation with this account?
The trend matters more than the level. A customer whose active users dropped from 12 to 7 in a month is more worrying than one who has always had 5.
How to make it useful
Check it against reality. Take the customers who churned last year. What did their signals look like three months before? Weight your score towards whatever actually showed up.
Keep it explainable. "Amber because active users fell 40% and the champion hasn't replied in three weeks" gets action. "Amber, score 62" gets ignored.
Cover every account. Health scores tend to get attention for big customers and none for the long tail. That's backwards. The small accounts are the ones nobody else is watching.
Tie it to an action. A score that changes colour and triggers nothing is decoration. Every drop should prompt something: a check-in, a helpful nudge, a fresh onboarding for a new user.
The bottom line
A health score is only as good as what it measures and what you do with it. Measure value, breadth, money and relationships. Watch the trend. Check it against who actually left. And make sure every account, not just the biggest, is being watched.
That's how Starla works. It reads usage, payments and conversations for every account, every day, explains why an account looks at risk and reaches out before it's too late. Curious what your accounts look like? Get a free churn audit.
Sources
- ChurnZero. Why your health score isn't predicting churn.
- Skok, D. (2018). Top Two Reasons for Churn. For Entrepreneurs.
