How to build a Gainsight health score your CS team actually trusts

Decide what the score is for

Before touching the scorecard, agree on the question it answers. Is it how likely is this customer to renew, or where should a CSM spend time this week? Those lead to different measures. A renewal risk score leans on usage, support and stakeholder signals. A prioritisation score might weigh contract value and upcoming renewal dates too. Pick one primary purpose.

Choose a small set of measures

Five to eight measures is a good starting point. More than that and nobody can explain why a customer is red. A balanced scorecard usually mixes a few groups:

• Product usage: logins, active users against licences, use of the features that matter most.
• Support: open ticket volume, escalations, time to resolve.
• Relationship: whether you have an engaged sponsor and champion, and recent executive contact.
• Customer voice: NPS or survey responses where you have them.
• CSM sentiment: a manual measure the CSM updates, with a required comment.
• Commercial: payment issues or downgrade requests.

Make every measure explainable

For each measure, write one sentence a CSM could say to a customer. For example, usage is red because fewer than half of your licences logged in last month. If you cannot write that sentence, the rule is too clever. Use Gainsight rules to calculate measures from your data, and keep thresholds simple at first: green, yellow and red bands are enough.

Weight it, then leave it alone for a quarter

Give more weight to the measures you believe predict churn best, usually usage and relationship. Do not overthink the first version. Launch it, let the team use it for a quarter, then adjust. Changing weights every week teaches people to ignore the score.

Fix the data before you blame the score

A lot of distrust comes from bad inputs rather than bad logic. Usage data that is a week late, support tickets matched to the wrong company, or CRM accounts that do not match Gainsight companies will all make the score wrong. Check that the company ID mapping between your CRM, product data and support tool is clean, and that data loads run on the schedule you think they do.

Keep CSM sentiment, but make it accountable

CSMs know things the data does not, like a new VP who wants to switch vendors. Keep a manual sentiment measure, but require a short comment whenever it changes, and show when it was last updated. A sentiment score that has not changed in six months is not a signal.

Check the score against what happened

Every quarter, pull the customers who churned or downgraded and look at what their score was 90 days before. Then look at the reds that renewed happily. If churned customers were mostly green, a measure is missing or weighted wrong. Gainsight also offers tools to help analyse scorecard measures against outcomes. This simple review is what turns a score into something people trust.

Connect the score to action

A score that does not trigger anything is just a colour. Use rules to create CTAs when a measure drops, attach a playbook for the most common risks, and review red accounts in your weekly team meeting. When CSMs see the score helping them act earlier, they start to trust it.

A quick checklist

1. Agree on one purpose for the score.
2. Pick five to eight explainable measures.
3. Set simple thresholds and sensible weights.
4. Clean the company ID mapping and data loads.
5. Keep CSM sentiment with required comments.
6. Review against churn and renewals every quarter.
7. Trigger CTAs and playbooks from score drops.