The template
Rate five signals from 1 (poor) to 3 (good), multiply each by its weight, and add them up. The result runs from 1.0 to 3.0. As a starting point, treat 2.5 and above as green, 1.8 to 2.4 as yellow, and below 1.8 as red, then adjust the thresholds against your own renewal history.
Health score: [Customer name] · Scored by: [name] · Date: [date]
| Signal | Weight | 1 | 2 | 3 | Rating |
|---|---|---|---|---|---|
| Results against their goals | 30% | Most goals off track | Mixed | Most goals on track | [1–3] |
| Sponsor engagement | 25% | No contact in 90 days | Replies, doesn't attend | Attended the last review | [1–3] |
| Usage by the people it was bought for | 20% | Falling | Flat | Steady or rising | [1–3] |
| Champion | 15% | None, or just left | One, not senior | Active, with influence | [1–3] |
| Support and escalations | 10% | Open escalation | Tickets rising | Quiet or improving | [1–3] |
Score: [sum of rating × weight] · Status: [Green / Yellow / Red]
What the score doesn't show: [anything you know that the five signals miss]
The weights are a starting point, not a finding. Outcomes and the sponsor carry the most because they're closest to the renewal decision. Change them when your own data says otherwise.
A worked example
The account below is fictional.
| Signal | Weight | Rating | Why | Points |
|---|---|---|---|---|
| Results against their goals | 30% | 1 | Two of three goals off track | 0.30 |
| Sponsor engagement | 25% | 2 | Replies to email, missed the last review | 0.50 |
| Usage | 20% | 1 | Weekly active reps fell from 71% to 52% | 0.20 |
| Champion | 15% | 1 | Left in April, not replaced | 0.15 |
| Support and escalations | 10% | 3 | Quiet | 0.30 |
Score: 1.45. Red.
Notice the last row. Support is quiet, and on a score built from tickets and logins alone this account might have looked fine for months. Quiet isn't always healthy. Sometimes nobody's filing tickets because nobody's using the product.
A score is a guess. Test it.
Here's the uncomfortable part. Most health scores are never checked against what happened.
Take every account that came up for renewal in the last year. What was its score 90 days before? If the accounts that left were mostly red and yellow, the score is doing its job. If half of them were green, it's measuring the wrong things, and a dashboard full of green is telling you very little.
Run that check once a year. Then change the signals and weights to match what you find.
What to do with a red or yellow score
The score tells you where to look. It can't tell you why. A drop in usage could be a reorganization, a champion who left, a missed training, or a competitor's trial. Each one needs a different response, and the score looks the same for all four.
So the next step is always a conversation. Call the champion. Bring your read of the cause to the sponsor and ask whether it matches what they see. The QBR questions page has the ones that surface renewal risk, and running a QBR with an unhappy customer covers the meeting itself.
Three mistakes to avoid
- Too many signals. Fifteen inputs produce a number nobody can explain. If you can't say why an account is yellow in one sentence, simplify.
- Only what's easy to count. Logins and tickets arrive automatically, so they dominate. The signals that predict renewal best usually need a person to judge them.
- Hiding the score from the customer. You don't have to share the number. But if you'd be embarrassed for them to see what's behind it, the scorecard you show them isn't honest either.
Get the QBR kit
Fifteen slides in meeting order, with notes on every slide: what goes there, what to ask and what to avoid. It also includes AI prompts that draft the deck from your notes, the prep timeline and a 60-minute run-of-show. PowerPoint, opens in Google Slides, $39.
Checkout opens soon. Get the free starter deck now, and the scorecard if you want to hear when the kit is ready.
Health score questions
What is a customer health score?
A customer health score is a number or color that summarizes how an account is doing, built from a handful of signals such as product usage, results against the customer's goals, sponsor engagement and support history. Teams use it to decide where to spend attention before renewal.
How do you calculate a customer health score?
Pick four to six signals, rate each on a simple scale, multiply each rating by a weight, and add the results. Then set thresholds for green, yellow and red. The template on this page uses five signals rated 1 to 3.
What signals should a health score include?
At least one from each of four kinds: outcomes (are they hitting their goals), usage (are the right people using it), relationship (is the sponsor engaged, is there a champion) and friction (support trend, open escalations).
What's a good customer health score?
There's no universal number. A score is good if it separates accounts that renew from accounts that don't. Check yours against the last year of renewals and churn.
How often should a health score be updated?
Usage and support signals can update automatically. Relationship and outcome signals need a person, and should be reviewed at least monthly and after every QBR.
Why do green accounts churn?
Usually because the score measures activity and the customer left over outcomes or people: a goal that was never met, a champion who left, a sponsor who stopped attending. Add those signals, and treat any score as a reason to ask questions.