In the AI era, CX and CS leaders are having to solve a new problem. Retention, expansion, and customer health are fast becoming board-level mandates, often owned by the CRO, and they are being scrutinized in an environment where teams are smaller, resources are tighter, and expectations for clear ROI are higher.
Welcome to Beyond The Click by Balboa Solutions. Today we're starting a series on how to lead CX in this AI-first, resource-constrained world. Each issue will cover one of the tactical topics that make up this theme, and today's is all about customer health scores, including:
📆 REGISTER HERE: Our next DX Roundtable is Feb 26th at Noon ET. Join Jay Nathan (Balboa) and Jenna McLaughlin (Pendo) for Leading CX in an AI-First, Resource-Constrained World. Jenna and Jay will share their takes on how AI is reshaping how teams protect and expand existing revenue before we open it up for group discussion.
🛫 Also, Pendomonium 2026 is just 7 weeks away. See the agenda here if you're still on the fence about going! The Balboa team will be there in force to share conversations about this topic and others. See you at the Balboa booth!
Let's dive in.
Most CS teams already do some version of health scoring, whether it's a quarterly "heat map" meeting or a dashboard with red/yellow/green indicators. But the uncomfortable truth is that these scores often become shelfware. Teams build them, roll them out, and then quietly stop trusting them. The number sits on a dashboard, and nobody cares.
Why does this happen? Because traditional health scores break under pressure.
Health scoring exists because the CS job isn’t designed to scale. When your CSMs have too many accounts, they triage. They need a consistent way to decide where to focus before renewals are "quietly decided" by customers who never raised their hand.
But manual scoring fails when your customer base expands. The issue is inconsistency: different CSMs weight different categories differently, turning risk assessment into "finger in the air" intuition instead of a truly data-driven, apples-to-apples comparison. And when you discover risk in customer conversations, it's often too late. Hearing "they stopped using feature X/Y/Z" in a meeting is a late-stage warning sign.
The best health scores share a few key traits:
Back-testing against customers who historically churned vs renewed is the key differentiator. You're testing the score against reality, not rolling out a hypothesis-based model and hoping it works.
Detailed adoption and engagement data (pages used, feature stickiness) helps you distinguish "logging in" from "getting value." CRM data still matters, but it's messy in real enterprises with parent-child account structures and data hygiene issues.
Signals and risk dynamics vary by customer size, industry, and deal type. A one-size-fits-all model misses the nuance.
Some data points just aren't natural for humans to conceptualize. Adoption improves when the view explains WHY an account is high risk in plain language.
Surfacing risk in Salesforce helps sellers and CSMs use it because they already live there and can start taking action.
Don't boil the ocean. Start with a simple, believable model (e.g. Healthy/Neutral/Unhealthy) and layer complexity over time.
Static customer health models go stale because maintenance is hard. And the conventional route of pulling in data science resources to build custom models is expensive, slow, and sometimes outdated by the time it ships.
Predictive churn modeling tools like Pendo Predict can help here. They enable you to make your customer risk modeling iterative and continuous, and to put it back in the hands of the people who know the customer best (your CS and CX teams) without requiring expensive data science personnel. And as outcomes occur (renewal, churn, expansion), teams can validate their predictions to build trust over time.
But however you build it, remember that the real churn signals are high-volume and multi-source. You need a system that can handle feature utilization changes, customer sentiment from conversations, and account context simultaneously. And you need it to adapt as your business evolves, without becoming another abandoned dashboard.
Did a friend forward you this article? Subscribe to Beyond The Click and get the latest on Pendo and Product Ops straight to your inbox every other Thursday.⚡