Signals arrive after the decision to leave has already been made.
Leaders want a signal they can act on, not a model they have to maintain.
One head of CS is torn: a formal score would simplify triage, but could be "a waste of time", so they lean towards a simple red, amber and green model. A post-sale advisor echoed the caution: "keep it small". In their experience, overbuilt health scores "die under their own weight". When they are too hard to produce, they do not get done at precisely the moment they are needed.
Timing was the bigger issue. Usage, billing, sentiment and meeting recaps were all called must-haves, yet they often fail to warn of churn soon enough. Once a customer gives notice, typically around 90 days before renewal, the same advisor estimated the chance of saving the account at "probably less than 10%". Their alternative is to design an expected customer progression from signature to value, then flag accounts that fall outside it.
"Customer is happy" is not a metric.
"You got to stop the bucket leaking first."
Leaders also stressed blending numbers with conversation. A payments CSM said quantitative data or sentiment alone each gives "an incomplete picture" of account health. A fractional CS leader added a maturity warning: below roughly 100 to 200 customers, teams should learn their signals manually before expecting a platform to discover them, and should track growth signals alongside churn signals.
A short list of leading indicators, measured against an expected journey, beats a complex score built on lagging ones.
Our free retention health check scores how early your signals arrive. No integrations required.
Check your health score →Renewal problems are usually onboarding problems that went unnoticed for a year.
Asked for a magic wand, one payments CSM wished they could "read my customers' minds". Customers often cannot articulate what they need until they see the product in action, and those needs shift during implementation. Getting that understanding right early prevents much of the friction later in the journey.
A post-sale advisor named the earliest churn signals they look for: a customer who cannot clearly state why they bought, or a delivery team that was never part of the sales process. A CS operations lead focused on expectation-setting: never overpromise, raise delays early, and build in buffer so the team can "look like a hero" rather than miss a promise. They also flagged how rotating territories and poor note-taking let difficult situations become the "next person's problem".
Several leaders stressed that the goal of onboarding is self-sufficiency. Rather than training every user, the strongest CSMs equip the client's own champion team, building internal expertise that deepens adoption and makes the account far harder to churn.
Gartner describes a "customer value gap": suppliers struggle to turn the promise of their value proposition into value the customer actually realises. Its recommendation starts with onboarding programmes that acclimatise customers to the partnership, followed by personalised learning that drives full adoption.⁵ That is exactly where our interviewees said the earliest signals live, and where most health scores are not yet looking.
Define what good looks like at day 30, 60 and 90, and flag deviations long before the renewal conversation.