By the time a customer's health score turns red, the renewal conversation has usually already gone wrong. Useful prediction fires earlier and names a cause.
- ›Why most health scores are lagging indicators
- ›The signals that lead rather than follow
- ›Making a score actionable
- ›The cross-department signals nobody watches
Why most scores lag
Common health scores combine product usage, support tickets, and NPS. All three are downstream: by the time usage drops or tickets spike, the customer has already had the experience that will decide the renewal.
They are also frequently confirmatory. A score that tells your account manager what they already sensed adds no information; the ones worth having tell you something you did not know.
Signals that lead
Engagement breadth, not depth. A single power user is fragile. The strongest predictor of churn is a shrinking number of engaged people, and it moves before total usage does.
Champion change. Your champion changed roles or left. This is the highest-value early warning in B2B and it is often visible before any usage metric moves.
Response latency. How quickly they reply compared to their own baseline. A customer who used to reply in hours now taking days is a signal, and it is measurable.
Unmet expectations from delivery. A project that slipped, a promised feature that did not arrive, a support issue that took too long. These predict renewal outcomes and live outside the CRM entirely.
Silence in a period that should be busy. Quiet during their peak season means they are not using you when they most need to.
Making a score actionable
A score is only useful with three attachments: the reason, the trend, and a suggested action. "62, down from 78, driven by two departed contacts and a delivery slip; suggested action: re-establish coverage with the new stakeholder before the December renewal" is worth reading. "62" is not.
Trend matters more than level. A stable 65 is often healthier than an 80 that fell from 95 last month.
Important: Do not let health scores replace conversations. They tell you which conversations to prioritize and what to ask about, which is a different and more useful job.
The cross-department signals
The strongest churn predictors typically live outside the CRM: delivery slipped, invoices went unpaid or were disputed, support escalated twice, the onboarding never finished. Each is held by a different team, and none of them is watching for renewal risk.
This is where a shared data layer changes the outcome rather than the reporting. When delivery, finance, support, and sales write to the same system, "which customers are at risk and why" becomes a query rather than a project. See what an autonomous business operating system is.
FAQ
How far ahead should a warning fire?
Long enough to act: at least one renewal cycle's worth of runway, which for annual contracts means 90 days or more.
What if we have too few customers for prediction?
Then use explicit rules based on the signals above rather than a model. Rules are transparent and work at small scale.
Who should own account health?
Whoever owns the renewal, with the signals visible to delivery and support too. Health owned solely by customer success misses the causes that originate elsewhere.
Brainis computes account health from sales, delivery, finance, and support data in one system, with reasons attached. See Revenue OS.
Sharing insights on business operations, AI, and modern team management.
Run your company on Brainis
All 11 Operating Systems on every plan, from $29 a month. No per-seat pricing — you pay for AI capacity, not headcount.
See pricing