Customer health scores explained (and why they lie)

Account K's health score was green, right up until procurement asked for the contract. That is not bad luck. It is how most health scores are built. Today, what goes into a customer health score, how to weight it, and why a score must explain itself. A health score is one number that sums up how likely an account is to stay and grow. Here is the framework. Four families of inputs, the same four layers as the last lesson. Words, themes, and tone in tickets, calls, and surveys. Relationships.

Is the champion still there, and how many people do you actually talk to? Behavior. Depth of usage, not just logins and whether core workflows are running. Commercial. Renewal date, payment history, expansion or contraction. Pick two or three inputs from each family, not everything you can measure. Next, waiting. Not every input deserves the same say. Look back at accounts that left and accounts that renewed. Which inputs moved first before the loss? Give those more weight. Write the weights down in plain words where anyone can read them.

Start simple. Equal weights are fine on day one, as long as you check them later, and set them per segment. A small team and a large enterprise do not show risk the same way. Now, the part most scores miss. A score must explain itself. 72 tells a customer success manager nothing. Down nine this month, the champion moved role, and reporting tickets keep repeating, tells them exactly who to call and what to bring. Every score should carry its top reasons and the evidence behind each one. Show the direction, too.

Up, down, or flat, and since when. A score without reasons gets argued with, then ignored. Let us rebuild account K's score. The old one read mostly usage, and logins held up because other teams at Account Key still used the product every day. Green. Rebuild it with the four families. Words, the reporting problem repeating and the tone turning. Amber. Relationships, champion gone. Red. Behavior, overall logins steady, reporting quiet. Amber. Commercial, renewal still ahead, neutral. Overall, amber with two reasons on top and it moves weeks before procurement ever writes.

Three traps to avoid. First, the usage only score. Usage is behavior, the third layer, so it moves late. Account Key's logins looked fine while its champion walked out. Second, too many inputs. 40 inputs means nobody can say why the score changed. Start with a handful. Third, never recalibrating. Your product and your customers change. Every quarter, check the score against who actually renewed and adjust. This week, write down every input in your current health score. Mark each one, words, relationships, behavior, or commercial. Then take the last five accounts you lost and ask what the score said three months before.

If it was green, you know which layer is missing. Customer intelligence explained is presented by Hyperorbit, a gentic customer intelligence. See a risk read that explains itself in the free churn prevention guide at hyperorbit.ai.