Gainsight first introduced the customer health score to the world of customer success in the early 2010s, and the goal was admirable: help CS teams use data to have more informed conversations with their customers.

However, almost every Customer Success Manager (CSM) has seen it: a customer marked “green” suddenly churns. Or an account flagged as “yellow” quietly expands. Customer health scores, once hailed as the cornerstone of the CSM’s account insights, are failing us.

It’s not because the idea was bad. Back then, it was the best we could do.

At the time, Gainsight (and other CSPs that followed) made the most of the technology infrastructure that was available. Product data was trapped across siloed systems. Real-time querying and statistical analysis didn’t exist. Scores had to be built using manual, limited, batch processes. 

The best we could do was generate a backward-looking snapshot to get a flat insight into our customer’s state, and maybe give us some direction for discussion points for our periodic customer check-in.

But the world has changed. The infrastructure has changed. And the job of CS has changed.

We now live in an environment where:

• Product, support, CRM, and transactional data are all available — in real-time — through centralized warehouses.
  
  

• Customers expect contextual, personalized guidance, only when they truly get value from an interaction — not generic quarterly check-ins.
  
  

• CSMs are expected to not only be trusted advisors, but also revenue engines that are responsible for retention and expansion.

• CSMs are stretched across dozens-to-hundreds of accounts (which have multiple users), have endless customer data to track to be in the know, and can’t afford to waste time chasing false alarms, or reacting after the fact.
  
  

The original health score wasn’t designed for this reality. It’s a thermometer. What we need now is a GPS.

What's Wrong with Health Scores?

Health scores fail at the exact moment they’re supposed to help — when timing and specificity matter most. Here’s why.

1. They’re Lagging and Late
   Health scores are built on data pipelines that aggregate multiple measures and metrics and move in batches. Some metrics are updated daily, others weekly or monthly. If a product usage drop is averaged into a monthly aggregate, you won’t see the impact until long after it started.
   
   Worse, most scores are calculated only once the last measure arrives (e.g. monthly consumption of certain credits or features) — which means the score missed dynamics of a whole month. That’s not helpful when you’re trying to catch a developing churn risk early or act on an upsell opportunity before it fades.

   
   

2. They’re Static and Untrusted
   Most health scores are based on fixed logic: a handful of measures, a set of static weights, and some business rules that were decided years ago — often during the initial CSP implementation. But your product evolves. Your personas and market shift. Your customer behavior changes. And yet, the scoring model stays frozen.
   
   And if the pipelines and transformations were broken, do we have a data team that is on it and alerts us that there’s an issue and it shouldn’t be used?

   
   

   So CSMs stop trusting it. What’s “green” for one team might look like “yellow” to another. Everyone starts ignoring the score — or worse, acting on it when they shouldn’t.

   
   

3. They’re Too Aggregated to Be Insightful
   To simplify a complex customer reality into a single number, health scores flatten everything: product usage, support activity, sentiment, commercial context. The result is an oversimplified picture that hides what’s actually going on.
   
   A customer might be increasing usage of one feature while disengaging from another that’s more critical. Or a new stakeholder might be added, but the score stays flat because the event doesn’t register in the weighted model.

   
   

   CSMs don’t need a number. They need to know what changed, why it matters, and what to do about it.

   
   

4. They Can’t See the Future Trajectory
   Health scores summarize the past — they don’t model the future.
   
   They don’t analyze whether an account is drifting, accelerating, or hitting a behavioral inflection point. They can’t compare a customer to its own historical baseline, similar peers, or expected benchmarks for its tier, industry, or lifecycle stage.

   
   

   Without that predictive context, you’re stuck reacting to symptoms instead of intercepting root causes.

   
   

   A CSM doesn’t just need to know what happened. They need to know where the account is heading — and what action to take next.

   
   

5. They’re Not Actionable
   Even when a score changes, it rarely helps you to infer what to do next. There’s no root cause. Not to mention a direct, data-driven (or statistical) recommendation. Just a shift from one color to another.
   
   This leaves CSMs stuck in detective mode — burning time trying to figure out why something changed and how to respond. By the time they do, the moment may have passed.

   
   

6. They Ignore Account-Specific Behavior Shifts
   Health scores rely on flat, point-in-time values — logins this week, NPS last month, number of open tickets. But these raw levels don’t tell us where the account is headed.
   
   To understand if a customer is on track for success — or veering off-course — we need to monitor changes in behavior that are unique to that account.

   
   

   Is engagement rising or falling compared to their usual baseline? Has usage of a success-critical feature suddenly dropped? Has a champion gone silent?

   
   

   Static measures can’t answer these questions. And global benchmarks miss the nuance. A login drop might be normal for one segment but alarming for another.
   
   

Only by detecting statistically meaningful change, correlated with churn risk or growth potential, can we understand what’s actually unfolding. Health scores weren’t built for that — so they miss it.

Why They Struggle: A System Not Built for Precision or Speed

These problems aren’t just implementation flaws — they’re symptoms of a model built for a different era. Health scores are fundamentally a business intelligence solution. Like a BI report, they summarize what already happened.

They don’t predict. They don’t adapt. And they certainly don’t guide you.

That’s fine for a management update or board deck. But it doesn’t work for a CSM managing hundreds of accounts, trying to intercept churn or capitalize on expansion before the window closes.

CS doesn’t need BI to proactively win. It needs decision intelligence — something fast, granular, and predictive. Something built for action, not just awareness of the past.

From Thermometers to GPS: What Comes Next

It’s time to replace the thermometer with a GPS.

A GPS doesn’t just tell you where you are. It tells you where you’re headed, flags problems and obstacles up ahead, and suggests the best path forward. That’s exactly what modern Customer Success teams need — and what today’s infrastructure finally makes possible.

The new model is powered by predictive signals. Here’s what it looks like:

• Real-Time Monitoring
  Cross-analyze granular, account-level, product usage behaviors, support patterns, stakeholder structural activity, and sentiment changes — continuously, in real-time, not in delayed batches.
  
  

• Account-Unique Change Detection
  Identify the specific shifts in behavior that statistically correlated with developing churn risk or expansion opportunity.
  
  

• Root Cause Analysis
  Instead of just flagging risk or opportunity, understand why it’s happening. Tie signals to causes: specific sudden struggles, success-related feature usage gaps, sudden lack of engagement, support friction, strategic misalignment.
  
  

• Next-Best Action
  Automatically surface what action should be taken — escalate, re-engage, offer training, propose a new use case — based on what’s worked in similar situations before, and for the customer’s objectives.
  
  

• Outcome Orientation
  Don’t just score risk. Solve it. Drive success with perfectly timed, contextual interventions.
  
  

A New Philosophy for CS

The old model says:

Measure what happened. React when you can.

The new model says:

Monitor continuously. Predict accurately. Act decisively, at the perfect time.

We don’t need better scores. We need better signals — and better timing.

If we want Customer Success to be truly proactive, then we have to stop flying blind.

See how your CS org can be proactive by contacting us today.