Health scores

Every Company in GainTrace carries a health score from 0 to 100 and a status: Healthy, At Risk, Churning, or Not scored. This article explains what feeds the score, where it appears, how to change the weighting, and what to expect when a company has little data.

​

How a score is computed

Health is a weighted average of up to six dimensions, each normalized onto a 0 to 100 scale, plus an adjustment from recent signals. It is a transparent calculation over your synced data, not an AI prediction: the same inputs always produce the same score, and Trace AI does not set the number. Each dimension needs a real data source behind it. A dimension with no data for a company is skipped, and the remaining weights renormalize, so the score reflects only what GainTrace actually knows. Weights are shares, not absolutes. If a company has data for only three dimensions, those three weights are rescaled to cover the whole score. Revenue is normalized on a logarithmic scale: a larger contract scores higher, but not proportionally higher, so one large company does not read as ten times healthier than a mid-sized one.

​

How signals adjust the score

After the weighted average, GainTrace folds in the company’s active signals from the last 30 days. Risk signals such as a usage drop, a champion leaving, or a failed payment subtract points; positive signals such as reaching first value or an expansion add points. An NPS detractor lowers the score and a promoter raises it. The total signal adjustment is capped between -30 and +20 points, so signals shift the score rather than replace it. Dismissing a signal removes its effect on the next scoring run, and signals older than 30 days age out on their own.

​

Read the statuses

A company with nothing to score shows Not scored rather than a fake zero. The score only exists when at least one dimension has real data.

​

Where health appears

• On the Companies list, each row shows the status in the Health column. The All health dropdown filters to Healthy, At Risk, or Churning, and the Sort by control in the Display menu can order the list by Health score. Clicking the Health column header sorts the list too.
• On a company’s record, the score and status appear in the metrics strip at the top of the page once the company has a score. A health trend chart covering the last 90 days sits further down on the Overview tab.
• On Boards, widgets such as average health and health movers read the same scores and their snapshot history.
• In Flows, add an Event trigger and set its event type to Health score change. It starts an automation when a company’s status changes or its score moves by 5 or more points.

​

Change the weights

Changing weights requires an Owner or Admin role. Other members can view companies’ scores but cannot edit the scoring profile. Weights do not need to add up to 100: each weight is a share of the total you set. One scoring profile applies to every company in the workspace.

​

When scores update

Scores recompute on an automatic cycle that runs several times a day, roughly every six hours. Each run refreshes the underlying inputs, recomputes every company’s score, and records a dated snapshot, which is what the trend charts draw from. Expect a change you make, such as a dismissed signal, a new NPS response, or saved weights, to be reflected within about six hours. A company’s trend chart needs at least two snapshots before it has anything to show.

​

What happens with sparse data

• A dimension with no data is skipped and the remaining weights renormalize, so a partial score is still a real score over what is known.
• A company with no scoreable data at all shows Not scored. GainTrace never fills the gap with an invented number.
• Adoption is skipped for every company until you define features under Features.
• Sentiment is skipped until GainTrace has computed the company’s advocacy score, which happens on the first scoring cycle after activity data syncs.
• Support is currently skipped for every company; its weight changes nothing until a support data source exists.

In a new workspace, connect a data source first: scores appear after the first scoring run that follows your first sync. The more sources you connect, the more dimensions participate and the fuller the score becomes.

​

FAQ

Is the health score AI-generated? No. It is a deterministic weighted calculation over your synced data. Trace AI drafts actions and explanations elsewhere in the product, but it does not decide the score. Does the health score predict churn? No. It summarizes the current evidence about a company; it is not a probability and carries no accuracy guarantee. Treat it as a guide for where to look first. Why did a score drop suddenly? A new risk signal, such as a failed payment or a champion leaving, can subtract several points in one run. Open the company’s record and check its recent signals for the cause. Can different segments use different scoring? Not today. The weights you set on the Health scoring page apply to every company in the workspace. Why does a company show Not scored? No dimension had data for it, and no CSM pulse was recorded, usually because its sources have not synced usable data yet. Once any dimension has data, the next scoring run produces a score.

​

Good to know

• A CSM pulse counts only where one has been recorded, and a pulse older than 90 days stops counting, so a stale gut call cannot quietly steer the score forever.
• Health is a guide, not a verdict. Use it to decide where to look first, then open the company for the full story.

​

Related articles