Your Health Score Is Green, but Your Account Is About to Churn. Here's Why.

Progress on the QBR slides looks on track; the account has a green health score, 85% on product adoption and no open escalations. The renewal is in three months and nobody is worried.

Suddenly, the customer's CFO restructures the organisation. The VP of Operations who championed the implementation, who'd fought for the budget, trained the team, and got the product live, is now gone. And nobody on the vendor side had noticed the VP stopped attending the monthly syncs six weeks earlier.

So the account churned. Green health score right up until the termination notice.

I've seen this happen at HubSpot, at Cisco, and now at ServiceNow. The post-mortem is always the same: the score didn't fail because the data was wrong. It failed because a single number was never designed to capture what actually precedes churn.

What precedes churn isn't a just red score. It's a pattern of quiet signals: specific, observable, almost always visible in your data 60, 90, 180 days before the conversation you don't want to have. Standard health models don't surface them.

It's other signals that often have a stronger effect on the renewal.

Here are six I've seen most consistently, and what to do when you spot each one.

The 6 Silent Churn Signals

1. Executive sponsor disengagement

Your champion stops attending meetings, maybe not dramatically, they just start sending delegates. In your CRM, meeting attendance looks fine because the account still shows activity, but the person who owns the internal budget justification has quietly stepped back.

What you need to start tracking: Flag any account where the named executive sponsor hasn't joined a meeting directly in 45 days. Don't wait for the CSM to notice; build the alert.

2. Champion exit and who replaced them

This is the one that kills the most renewals and almost no health model catches it. Your champion leaves the company. A new decision-maker comes in from outside - often from a competitor, sometimes from a company where they used a different tool. They have no relationship with you, no institutional investment in the product, and they're three months into a new role where proving their own value matters.

At ServiceNow, we learned to treat any champion departure as a commercial trigger, not just a relationship task. The new stakeholder's background matters. Where did they come from? What tools did they use there? Have they already mentioned anything about "evaluating the current stack"? That's not curiosity; it's due diligence for a decision they've already started making.

What to do: The moment a champion exit is logged, run a stakeholder mapping exercise and assign an exec-to-exec introduction within 30 days. If you don't have exec coverage at that account, you will have a tricky renewal when the time comes.

3. Adoption plateau after onboarding

Usage hits a ceiling at month three or six and stays flat. Standard models read this as stable. What it really means is this: the customer implemented enough to justify the initial purchase, but never reached the use cases that would justify renewal - meaning the intended value. Logins might be consistent; they're just not doing anything meaningful.

What you need to start tracking: Build a depth-of-use metric alongside your adoption rate. If the same three features account for 90% of activity at month six, that account is a contraction risk regardless of what the score says.

4. Support ticket topic shift

Watch what customers are asking, not just how many tickets they raise. When tickets shift from "how do I do X" to "why doesn't X work" or "we need X exported," the product is hitting its ceiling for their use case or something was done wrong. In isolation, a spike in support volume looks like engagement; in reality, it's a customer preparing the business case for switching.

What you need to start tracking: Tag support tickets by intent: how-to, product limitation, data export. If limitation and export tags are trending up in an account 90+ days before renewal, flag it for a proactive conversation.

5. Silence after a product change

A major feature launch or pricing change goes out. Some customers light up; they ask questions, they test, they engage. Others go quiet. A customer who doesn't react to something you expected them to care about has usually already started reviewing their options.

What to do: For any product change that affects your top 20% of accounts by ARR, build a 14-day response check into the playbook. No response is data - treat it as one.

6. Stakeholder map erosion

Six months ago, five people at this customer were actively engaged with your product. Now you have two: both in operational roles, neither in a position to make or influence a renewal decision. No one left the company, they just stopped being involved. By the time the renewal conversation starts, there's no internal champion to run it on your behalf.

What you need to start tracking: Map stakeholder engagement by seniority, not just headcount. Breadth of contacts isn't the same as depth of coverage. An account that narrows to operational-only engagement is already at risk.

None of these signals are exotic. Every experienced CSM recognises them the moment they're named. The problem isn't awareness; it's that standard health models don't track them systematically, so they exist only in the CSM's head - invisible to the forecast, invisible to leadership, invisible until the termination notice.

The health score had one job: tell you which accounts need attention. For years it's been doing something else - telling you which accounts looked healthy when someone last updated the model. According to the 2026 State of Customer Success report, 60.7% of CS teams review their health scores on an ad-hoc basis or never. That's not a tooling problem, but a design problem.

The leaders who survive the next efficiency wave won't be the ones with the most sophisticated score. They'll be the ones who stopped using a number to do a system's job.

The six signals above are your team's job to catch. But there's a prior question that only you, as a leader, can answer: does your current health model even give them the infrastructure to catch them?

Pull up your health score logic this week. Ask three things:

1. Is executive sponsor engagement tracked as a named, dated field or inferred from general account activity?
2. Does a champion exit trigger a commercial escalation, or just a CRM update?
3. When was the model last rebuilt against actual churn data from your portfolio?

If you can't answer all three cleanly, you don't have a health system. You have a dashboard that's giving your board false confidence and giving your team false cover.

That's the conversation to have with your CRO before the next QBR, not after the next churn notice.

Which of these six signals would your current health model catch automatically and which one would only surface because a CSM happened to mention it in a call?

— Iliyana

CS in the AI Era is a bi-weekly newsletter for Customer Success leaders in European SaaS. If someone sent this to you and you want to receive it directly, subscribe here. Or if you know a CS leader who should read it? Forward it.