A composite score, usually 0–100, that combines weighted product-usage, engagement, and billing signals to predict how likely each customer is to renew, expand, or churn.
Health score = Σ (signal score × signal weight), where each signal is scored 0–100 and the weights sum to 100%
An account scored on four weighted signals: product usage 80/100 (weight 40%), engagement 60/100 (weight 30%), support health 90/100 (weight 20%), and billing health 100/100 — no failed payments or downgrades (weight 10%).
(80 × 0.40) + (60 × 0.30) + (90 × 0.20) + (100 × 0.10) = 32 + 18 + 18 + 10 = 78/100 — top band, a candidate for an expansion conversation rather than a rescue call
A customer health score is a leading indicator where churn rate is a lagging one. Rather than reporting who already cancelled, it estimates who is drifting towards cancellation by rolling four to six signals into one number per account: product usage depth and frequency, engagement such as logins and support interactions, and billing signals like failed payments and downgrades. The output is a triage list, not a precise probability.
The weights should come from your own churn history, not gut feel. Review the accounts that cancelled over the last 6–12 months and ask which signals moved first. If churned customers went quiet in the product before leaving, usage deserves the heaviest weight; if downgrades preceded most cancellations, weight billing up. A common starting split is 40% usage, 30% engagement, 30% support and billing, recalibrated quarterly.
The most common mistake is overweighting flattering signals and ignoring billing ones. A customer can log in daily right up to the day they cancel, and promoters churn too. A failed payment, a seat reduction, or a switch from annual to monthly billing is among the strongest churn predictors you have, and it already sits in your Stripe data. The second mistake is stuffing in fifteen metrics: past 4–6 inputs the score gets noisier, not smarter.
Most B2B SaaS teams bucket scores into three bands — for example 0–40 (intervene now), 41–70 (monitor and engage), and 71–100 (stable, potential expansion). The bands matter more than the raw number: the score's entire job is deciding which accounts get proactive attention this week. And if you run on Stripe, billing data alone gives you a minimum viable health score before you touch product analytics.
By the time a cancellation shows up in Stripe, the decision was made weeks earlier — churn rate can only tell you what you already lost. A health score moves retention from reactive to proactive: it tells you which accounts to call this week, before the cancellation email arrives. For a founder with a few hundred customers, every saved logo is a visible chunk of MRR, and a pre-mortem is far cheaper than a post-mortem.
ChurnZero's 2024 Customer Success Leadership Study found only 34% of SaaS customer success teams had fully implemented a health score visible to all customer-facing roles; another 34% had one partially in place, and 20% planned to build one within 12 months.
There is no universal number — a good score is calibrated to your own churn history. On a 0–100 scale, most B2B SaaS teams treat roughly 71–100 as healthy, 41–70 as needs attention, and 0–40 as critical risk. The test of a good model is not the threshold but whether accounts that later churned actually scored low beforehand.
Pick 4–6 signals that historically preceded churn or expansion — product usage, engagement, support activity, and billing events — score each from 0–100, assign each a weight summing to 100%, and take the weighted sum. Review churned accounts quarterly and shift the weights towards the signals that actually moved first.
The strongest inputs are product usage depth and frequency, engagement signals like logins and feature adoption, customer feedback such as NPS, and billing signals — failed payments, downgrades, and contraction. Billing signals are the most commonly under-weighted despite being among the best churn predictors, and they are already in your Stripe data.
NPS is a single survey-based input: it measures stated sentiment at one moment. A health score is a composite of observed behaviour — usage, engagement, and billing — with NPS at most one weighted ingredient. Behaviour predicts churn better than sentiment; promoters cancel too. See our NPS definition.
Connect your Stripe account and see your real MRR, churn, and LTV in real time — on desktop and mobile.
No credit card required · Connect Stripe in 1 click
No credit card required. Connect Stripe in 1 click.
We're building iOS and Android apps that'll bring your metrics everywhere.
Our mobile apps are currently in active development. Follow us on social media for updates and be the first to know when they launch.