One number that tells you which customers are leaving — before they know it themselves
Ask an account manager how an account is doing and you get a paragraph. A customer health score compresses that paragraph into a single 0–100 number that can be sorted, trended and turned into alerts. Here is what goes into it, how to weight it, and how to read it without fooling yourself.
What a customer health score is
A customer health score is a weighted composite of the signals that predict whether a customer stays, grows or leaves. It is a leading indicator: it moves before the cancellation does, which is its entire reason for existing. A good score lets you look at a list of two hundred accounts and know in seconds which are thriving, which need attention, and which are at risk of churning before you next speak to them.
The score is deliberately reductive — that is its job. One number cannot capture the nuance of a relationship, but it can capture the trend, and the trend is what triggers the conversation that recovers the nuance. Teams that resist scores because "customers are more than numbers" usually end up with no systematic view at all, which is a worse outcome for the customers.
What goes into the score
Most useful scores mix five signal families. Engagement and usage — logins, active users, core-feature adoption — is the strongest predictor of retention for most products, because a customer who stopped using the product has already left in every way except the paperwork. Satisfaction and sentiment add what customers say in surveys, reviews and support conversations, catching shifts between surveys that usage alone misses.
Support experience measures open tickets, ticket age and repeat contacts about the same issue — the friction your team controls directly. Commercial relationship captures payment history, plan level and time to renewal. Finally, recency weights all of it: a six-month-old satisfaction score should not carry the weight of last week’s behaviour, so signals decay as they age.
How to calculate and read it
The standard method is simple: score each signal from 0 to 100, assign weights that total 100%, and multiply. For example, engagement at 80 weighted 30% contributes 24 points, satisfaction at 70 weighted 30% contributes 21, support at 90 weighted 20% contributes 18, and so on. Sum the contributions and an account scoring 78 across the board lands at roughly 78 — with the individual components telling you which lever to pull if it drops.
Read the result in bands and act on the band. Below 40 is at risk: assign an owner and a conversation this week. 40–69 needs attention: agree a fix for the specific friction. 70–84 is healthy — watch for drift, because a declining trend inside this band matters more than the level. 85 and above is thriving: the accounts to ask for expansion, testimonials and referrals while the goodwill is current. And publish the recipe — a score nobody understands is a score nobody trusts.
ClientTell keeps every customer's health score live
No spreadsheet, no monthly catch-up — the score updates as the signals arrive.
- Every customer scored 0–100 from satisfaction, engagement, sentiment, support issues and complaint frequency, with the recipe visible
- The whole book sorted by risk so the bottom 20% — where most churn lives — is always one glance away
- Alerts the moment an account crosses a threshold, with the feedback that moved it attached for context
Frequently asked questions
How is a customer health score different from NPS or CSAT?
What weights should I use for my business?
How do I calculate a customer health score?
Related tools & guides
Know which customers need you — before they leave.
ClientTell scores every account automatically from the feedback you already collect. Free to start.