What is a Customer Health Score?

A customer health score is a single rating that combines usage, support, sentiment and commercial signals for an account, so customer success teams can see which customers are likely to renew, expand or churn before the renewal date.

At a glance

• Gainsight describes it as a predictive metric for the likelihood of renewal, growth or churn.
• Scores are shown as red, yellow and green, as letters, or on a 0 to 100 scale.
• Most models use four to six inputs, each with a weight.
• Different segments need different models. A new account and a three-year customer look healthy in different ways.
• The weights should come from your own renewal and churn history, then get checked against it.

How a health score is built

Pick the signals that tell you whether a customer is getting value. Gainsight groups them into five kinds: behavior (feature adoption, logins), support (ticket volume, resolution time), relationship (executive engagement, contact with the CS team), financial (renewals, invoice history, upsells) and feedback (NPS, CSAT).

Each signal gets its own score, each score gets a weight, and the weighted scores add up to the account’s rating. In Gainsight’s software each input is called a measure, measures can be grouped, and the scorecard’s overall score is set from them automatically or by hand for subjective ones like sentiment.

A worked health score example

Gainsight’s guide gives a sample weighting: usage 40%, support trends 25%, sentiment 20% and executive engagement 15%, with accounts grouped as healthy (71 to 100), at risk (31 to 70) or critical (0 to 30). Treat that as a starting template, not a standard.

Example (illustrative): an account scores 80 on usage, 50 on support, 60 on sentiment and 40 on executive engagement. The weighted total is 32 + 12.5 + 12 + 6, or 62.5, which puts it in the at-risk band even though usage looks fine. The support and executive scores are what to look at first.

Common health score mistakes

• Weighting what’s easy to collect. Login counts are easy to pull. If they don’t predict churn in your data, they shouldn’t carry the score.
• Relying on CSM opinion. Manual sentiment has a place, but a score built mostly on gut feel changes with whoever owns the account.
• One model for every customer. Gainsight suggests weighting onboarding milestones for new accounts and ROI, product depth and upsell activity for mature ones.
• Tracking too many signals. Fifteen inputs make a number nobody can explain. Four to six that you can defend work better.
• Building it and never acting on it. A red score needs an owner and a next step, or it’s just another dashboard.
• Never recalibrating. When the product or customer mix changes, check last year’s model against who actually churned.

How a health score shows up in a revenue system

A health score only helps if someone works the list it produces: calls the red accounts, fixes the onboarding gap, raises expansion with the green ones. At AtoB, we run the customer-facing side alongside sales, covering support, collections, onboarding and QA. Read how the AtoB customer team runs.

Terms that sit next to customer health score

A falling score is the early warning for logo churn and churn rate, and a rising one is often where expansion revenue starts. Scores for new accounts lean on onboarding and time to value, and the score usually feeds the agenda for a QBR.