Customer Health Score Template: A Practical B2B Example

Choose a group whose customers buy the product for a similar job. A new customer still setting up the product should not be scored against a mature customer with an established process. Keep those groups separate before deciding what counts as healthy.

Write down the result the customer wants. For a reporting product, that might be a team using the report in its weekly planning meeting. Opening the dashboard is easier to count, but it is weaker evidence of value. Ask which event would convince both the customer and the CSM that progress is real.

HubSpot’s customer health score guide describes combining usage, engagement, support and feedback signals. The weights and thresholds below are our illustrative design, not HubSpot defaults.

Use four dimensions. For each one, define what 0, 50 and 100 mean before a CSM enters a score. Keep the raw evidence, its date and its source beside the score. This makes a disagreement useful: the team can discuss the evidence instead of arguing about a color.

These weights deliberately give customer outcomes the largest share. A different product may need a different balance. Do not use the same weights across all segments just because a single dashboard is easier to maintain.

Imagine the account scores 80 for outcomes, 60 for adoption, 50 for relationship and 90 for low friction. Its weighted score is 80 × 0.35 + 60 × 0.30 + 50 × 0.20 + 90 × 0.15 = 69.5. This is an arithmetic result, not a 69.5% chance of renewal.

The next step is visible underneath it: the relationship dimension needs a check, and adoption needs a clearer explanation. A CSM should confirm whether the customer still has an active project owner and whether the intended users can complete the core workflow.

You might initially review scores below 70. Treat that threshold as a testable operating choice. A critical unresolved issue can also require attention even when the average score is high.

An old usage sync and an absent survey are data problems. They do not prove that the customer is happy or unhappy. Add a coverage field that shows which dimensions are current, and define how long each source remains useful.

For example, a weekly usage measure could need a fresh weekly import, while an agreed success milestone may remain valid for a longer period. These are source-specific rules. If the evidence is stale, assign the work to fix it before sending an automated message about supposed customer risk.

In HubSpot, keep definitions and ownership explicit. If you use native health scoring, verify the features available in your subscription. If you use properties and an external calculation, preserve the same evidence trail. Neither setup removes the need for judgment.

Review a sample of accounts with different scores and record whether the recommended action made sense. Track missed issues, unnecessary escalations and accounts that could not be scored. When outcome data becomes available, compare results within the same segment and time window.

Change one part of the model at a time and keep a version date. Otherwise, a rise in average health may simply mean you changed the weights. The goal is a more useful account review, not a greener chart.

Use the customer success playbook template to define the response. For the separate question of which HubSpot renewals need follow-up, Renewal Radar resolves supported timing and operational evidence. It does not calculate this general health score.