Software Growth

Customer health score

A customer health score combines usage, support, payment and engagement signals into one score showing how likely an account is to renew or churn.

A customer health score is a single indicator per account, such as 0 to 100 or red, yellow and green, that summarizes how well that customer is doing with your product. The aim is to spot trouble early and to find accounts ready to grow. It is mostly used by customer success teams, and it works for a founder with 50 accounts as well as for a team with 5,000.

What goes into it

Gainsight, which popularized the practice, defines health as how customers interact with your product, how deeply they use it and how often, in its post on redesigning the health score. Common inputs:

  • Product usage: logins, active users per account, time in core workflows.
  • Depth: how many of the key features are used.
  • Support: ticket volume, unresolved issues, satisfaction replies.
  • Commercial: payment problems, contract end date, seat utilization.
  • Sentiment: NPS or survey replies.

How to calculate one

Score each input from 0 to 100, then weight them.

Take four inputs. Usage scores 80 at weight 0.4, feature adoption 50 at 0.3, support 90 at 0.2 and payment history 100 at 0.1. The score is 32 + 15 + 18 + 10 = 75. Mark 70 and above green, 40 to 69 yellow and below 40 red. The weights and thresholds here are an illustration, not a standard.

How to make it useful

  • Test it against history. Take accounts that churned last year and check what their score would have been three months before. If the scores were high, the model is wrong, not the customers.
  • Automate it. Gainsight lists replacing manual updates with automated measurement as a main fix, so that a customer turning from green to red is noticed.
  • Use different models by stage. A customer in week two needs different signals than one 18 months in. Gainsight also criticizes the old approach of simple traffic lights.
  • Attach an action to each color. A red score that no one acts on has no value.

Health score versus churn prediction

A health score is a hand-built summary. Churn prediction tries to estimate a probability from past data. Start with the first and move to the second when you have enough churn events to learn from.

For small SaaS

You do not need software. A spreadsheet with last login date, active users and plan, sorted by days since last activity, is already a health score. Every Monday, email the five accounts that went quiet with a short, human message. The pattern of who replies teaches you more than a weighted formula.

Sources

Back to the SaaS glossary