What it means
The score is built by choosing a handful of inputs, scoring each one on a common scale, weighting them by how strongly they predict retention, and adding them together. Accounts are then usually placed into bands, commonly green, amber and red, so that action can be triggered automatically rather than depending on somebody noticing.
The bands matter more than the exact number, because a score of 68 and a score of 71 rarely justify different responses. It matters because retention is where most of the profit sits in a recurring revenue business.
Winning a replacement customer typically costs several times more than keeping an existing one, so anything that gives early warning of a departure has direct financial value. Health scores turn scattered signals across product, finance and support systems into one prioritised list.
In use, customer success teams review the red and falling-amber accounts weekly, finance uses the aggregate distribution to sharpen renewal forecasts, and leadership tracks the share of revenue sitting in each band as a risk measure. Weighting by contract value is common, since ten small red accounts may matter less than one large one.
The essential nuance is that a health score is a hypothesis, not a fact. Its usefulness depends entirely on whether the chosen inputs and weights actually predicted churn historically, so a team should test the model against the last twelve months of renewals and cancellations and adjust it.
A score that never gets validated tends to drift into false comfort. Human judgement should also sit alongside the model.
An account manager who knows the client's sponsor is leaving next month holds information no system captures, so most mature processes allow a manual override with a recorded reason. That reason is itself useful data for improving the model later.
In practice
Real-world examples.
Example
A payroll software company weights payment behaviour heavily after finding that late payers churned twice as often as prompt ones. Accounts falling below 50 are routed to a save programme that includes a commercial review as well as a training offer.
Example
A cloud storage provider tracks the percentage of annual recurring revenue in the red band and reports it to the board each month. When the figure rises from 6% to 14%, leadership funds two additional customer success hires rather than waiting for the renewals to fail.
Example
A recruitment platform notices that health scores drop sharply whenever a client's main user changes. It adds a champion-change flag as an explicit component, which improves the model's ability to predict cancellations a quarter in advance.
Think of it
“Health score predicts how well a customer relationship is doing-green, yellow, or red status.
Formula
Calculation
Customer Health Score = Sum of (Component Score x Component Weight), where the weights total 100%
A software provider uses five components. Product usage carries a 35% weight, support experience 20%, relationship and engagement 20%, payment behaviour 15%, and adoption of key features 10%. Each component is scored from 0 to 100.
One customer, a regional retailer on a $60,000 annual contract, scores as follows: product usage 70, support experience 90, relationship and engagement 60, payment behaviour 100, key feature adoption 40.
Weighted parts:
Product usage: 70 x 0.35 = 24.5
Support experience: 90 x 0.20 = 18.0
Relationship and engagement: 60 x 0.20 = 12.0
Payment behaviour: 100 x 0.15 = 15.0
Key feature adoption: 40 x 0.10 = 4.0
Customer Health Score = 24.5 + 18.0 + 12.0 + 15.0 + 4.0 = 73.5 out of 100
At 73.5 the account sits in the green band, which starts at 70. The weak spot is feature adoption at 40, so despite the comfortable headline score the account team should book an enablement session on the unused modules well before renewal.Case study
Seen in the real world.
This case is illustrative and the company is fictional. Ashgrove Cloud, an invented data platform, had a health score built by its product team using five usage signals, and it showed 88% of accounts in green throughout the year. Renewals then came in at 74%, and the board wanted to know why the score had been so quiet.
The review found the model measured only what the product could see. It had no input for payment delays, none for support ticket sentiment, and none for whether the original buyer still worked at the client, and all three had been visible warning signs in most of the accounts that left. The score was not wrong, it was simply blind to most of the reasons customers actually cancel.
Ashgrove rebuilt the model with usage weighted at 40% and the remaining 60% split across support, payment, relationship strength and executive sponsorship, then back-tested it against the previous two years. In the illustrative retest, the revised score would have flagged 11 of the 14 accounts that churned at least one quarter before they gave notice, and the team restructured its weekly review around the red list.
Watch out
Common mistakes.
- Building a health score only from product usage data. Payment behaviour, support experience and the strength of the sponsor relationship often predict cancellation better than logins do.
- Trusting the score without back-testing it. Until the model has been checked against accounts that actually churned, the weights are guesses dressed up as a number.
- Treating every red account the same. Weighting by contract value ensures the team spends its limited time where the revenue at risk is largest.
Questions
People also ask.
How is a health score different from an engagement score?
Engagement scores focus on how much a customer uses the product, while health scores are broader and include commercial and relationship signals that predict renewal.
How often should health scores be updated?
Weekly is common for software businesses, and the change in score matters more than the level, since a sharp fall from 85 to 60 is more urgent than an account that has sat quietly at 62 all year.
Should the score be shared with the customer?
Generally no, because the model is internal and often contains subjective judgements, though the underlying facts such as low adoption of a paid module are well worth discussing openly.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%