Client Health Score: How to Predict Renewals Before They Slip

Most service businesses find out a client is leaving the day the cancellation email lands. By then the decision was made weeks ago, in a meeting you were not in. A client health score fixes that. It turns the signals your team already sees, slipping results, skipped calls, slow replies, a new boss on the client side, into one number you review weekly and act on before the renewal conversation.

The cost of not having one is bigger than it looks. On a 40-client book paying $5,000 a month, the difference between 6 percent and 4 percent monthly churn is roughly 10 clients a year. At a 10-month average lifetime, that is close to $480,000 in revenue that walked out while everyone was busy delivering.

I have seen teams build 20-field dashboards nobody updates, and founders run retention on gut feel. Neither works at scale. Here is the model we use: four to six inputs, a 100-point score, a playbook per color, a weekly review, and a quarterly check against real churn.

What a Health Score Actually Has to Do

A health score has one job: predict which clients will renew early enough to change the outcome, which means flagging risk 30 to 60 days before the decision. Three rules keep it useful.

Every input must be observable weekly without a research project. Every input must be something your team can respond to. And every score is owned by one named person, usually the account manager or CSM, who updates it and answers for it.

The Six Inputs That Predict Client Renewals

Pick four to six of these. More than six and the score becomes a data entry chore nobody updates by week three. Each input includes the scoring bands we use.

1. Results vs Goal (30 points)

Results matter most, because they are what the client pays for. Set one primary goal at onboarding, like booked calls per month or revenue added, and track pace. Proactive client reporting keeps that goal visible on both sides, so nobody is surprised when the score moves.

At or above goal pace scores 30. Between 80 and 99 percent scores 20. Between 60 and 79 percent scores 10. Below 60 percent scores 0.

2. Engagement (20 points)

Is the client still doing their part? Look at call attendance and client-side tasks like approvals, assets, and access over the last 30 days.

Full attendance with tasks on time scores 20. One missed or rescheduled call, or one task stalled over a week, scores 10. Two or more misses, or tasks stalled two weeks or longer, scores 0.

3. Responsiveness (15 points)

Measure reply time against the client's own baseline. A founder who always took two days to reply is normal. One who used to reply within the hour and now takes three days is a signal.

Normal for them scores 15. Noticeably slower scores 8. Needing two or more chases scores 0.

4. Sentiment (15 points)

The account owner logs tone after every call, backed by one monthly pulse question: "On a scale of 1 to 10, how confident are you that this is working?"

Proactive praise or a 9 to 10 scores 15. Neutral tone or a 7 to 8 scores 8. Questions about value or pricing, or a 6 or below, score 0. Cap sentiment at 15 points so gut feel cannot override hard data.

5. Payment Behavior (10 points)

Paid on time scores 10. A failed card that clears on retry, or one to seven days late, scores 5. Eight or more days late, a dispute, or a request to pause billing scores 0. Payment friction is often where internal doubt shows up first.

6. Stakeholder Changes (10 points)

The person who bought from you is your champion. If they leave or a new executive arrives with a mandate to cut vendors, your contract is suddenly up for review.

Same champion and decision maker scores 10. A new person added, or the champion changing roles, scores 5. Champion gone, new leadership, layoffs, or an acquisition scores 0.

Building a Simple 100-Point Client Health Score Model

Add the six inputs and you have a score out of 100. The weighting reflects a simple truth: clients forgive slow replies when results are strong, and rarely forgive weak results no matter how pleasant the calls are. If you only track four inputs, keep results at 30 to 40 percent of the total and spread the rest.

Then add overrides, because averages hide cliffs. Any of these forces a client to red regardless of score:

The champion leaves the company or the account.
Payment is 15 or more days late.
The client mentions cancelling, pausing, a budget review, or another provider.
Results have been below 60 percent of goal for two consecutive months.

Finally, add a trend rule. A client who drops 15 or more points in two weeks gets treated one color worse than their score says. A 72 that was an 89 two weeks ago is more dangerous than a stable 65.

Red, Yellow, and Green Thresholds With a Playbook for Each

Green is 75 to 100, yellow is 50 to 74, red is below 50. A score is only worth tracking if every color triggers a specific action with a deadline.

Green (75 to 100): Grow the Account

Green clients are where expansion, referrals, and testimonials come from. Within the month, the account owner asks for a referral, requests a case study, or proposes the next logical service. This is the moment for upsells and cross-sells that increase client lifetime value, because the client already trusts the results.

Yellow (50 to 74): Diagnose and Deliver Value

Yellow means drift, and the drivers are usually the same early warning signs of churn every operator learns the hard way. Within 72 hours, the account owner finds the lowest-scoring input and acts on that cause. No generic check-ins.

Lead with a fix or quick win, then ask a direct question: "Booked calls are at 78 percent of target this month. Here is the change we are making this week. Has anything shifted on your side I should know about?"

Red (Below 50): Escalate Within 24 Hours

Red gets a senior person. Within 24 hours, the head of customer success or the founder calls the client's decision maker, names the problem, and commits to a written recovery plan: what changes, by when, and what result to expect within 14 days. Open with ownership: "I am not happy with where your results are, and I would rather tell you that than wait for you to tell me." The client stays on the review agenda until green, or stable yellow for two weeks.

Run a 30-Minute Weekly Health Score Review

Account owners refresh every score by Monday at noon, about two minutes per client. The review runs the same day every week and fits inside a one-page weekly KPI scorecard.

Last week's actions (5 minutes). Did every assigned action happen? If not, why not.
Reds (10 minutes). Each red client's root cause, recovery plan, and owner.
Biggest movers (10 minutes). Sort by change in score, not absolute score. The client who dropped 20 points matters more than one sitting quietly at 70.
Assign actions (5 minutes). Every flagged client leaves with one owner, one action, and one due date.

Greens get skipped unless they are expansion candidates. When each customer success manager owns 15 to 25 clients, this review shows leadership the whole portfolio without sitting in every call.

Worked Example: Scoring One Account in a 40-Client Book

Take a 40-client book averaging $5,000 a month. The first week of scoring shows 26 green, 9 yellow, and 5 red, when the founder's gut said everyone was fine. Now zoom in on one client paying $6,000 a month, renewing in 45 days:

Results: 33 booked calls against a goal of 40, or 82 percent. 20 of 30.
Engagement: rescheduled one call. 10 of 20.
Responsiveness: used to reply same day, now two to three days. 8 of 15.
Sentiment: their team asked what they are getting for the spend. 0 of 15.
Payment: on time. 10 of 10.
Stakeholders: same champion on file. 10 of 10.

Total: 58, yellow. But two weeks ago this client was an 81, so the 23-point drop triggers the trend rule and red treatment. The head of CS calls within 24 hours and learns the founder hired a marketing lead who is auditing every vendor, a stakeholder change nobody had logged. The fix: a 14-day plan to get booked calls back to goal pace, plus a results walkthrough with the new marketing lead.

With eight more expected months, that client is worth $48,000. Save two like that per quarter and the system pays for itself.

Validate Your Client Health Score Against Actual Churn

A health score is a hypothesis until tested against what clients actually did. Most teams skip this step, and it is what makes the score trustworthy.

Keep history. Log every weekly score with a date instead of overwriting it, inside your single source of truth for business data.
Pull outcomes quarterly. List every client who renewed or churned in the last 90 to 180 days.
Check the catch rate. What percentage of churned clients were yellow or red at least 30 days before leaving? Aim for 80 percent or higher.
Check false alarms. If more than half of red clients renewed without any intervention, your thresholds are too tight.
Post-mortem every surprise. Any client who was green 30 days before churning gets a written review. Find the missed signal and decide whether it becomes an input or an override.

Adjust weights in 5-point increments, one change per quarter, so you can tell what helped. Wait for at least 10 churn events before reweighting. With fewer, you are tuning to noise.

A typical first validation: 11 clients churned over two quarters, and 8 were yellow or red 30 days out, a 73 percent catch rate. Two of the three surprises had a stakeholder change logged late, so champion departure became an override and every monthly call now asks: "Has anything changed on your team or in your priorities?"

Common Client Health Score Mistakes

Too many inputs. A 12-factor model looks rigorous and gets abandoned in a month. Six inputs updated weekly beat twelve updated never.
Scoring activity instead of outcomes. Emails sent and hours logged measure your effort, not the client's experience.
No playbook attached. A score that triggers no named action with a deadline is a colorful spreadsheet.
Owners grading their own homework. People want their clients to look green. Have a manager spot-check five random greens each month against raw data.
Scoring new clients on results too early. For the first 60 to 90 days, score results on the milestones in your client onboarding SOP instead of goal pace.

Frequently Asked Questions

What is a client health score?

A client health score is a single number, usually out of 100, that combines signals like results, engagement, responsiveness, payment, sentiment, and stakeholder stability to predict whether a client will renew. Its value comes from the action each score range triggers, not the number itself.

What metrics should be in a customer health score for a service business?

The strongest inputs are results against the client's goal, engagement on calls and tasks, response time versus their own baseline, sentiment, payment behavior, and changes in who owns the relationship. Weight results heaviest, around 30 percent. Skip vanity metrics like portal logins unless they track with renewals in your data.

How often should client health scores be updated?

Weekly. A client can go from fine to gone inside a single month, so monthly updates are too slow. Weekly scoring takes about two minutes per client and gives you a trend line, which is often more predictive than the score itself.

What is a good client health score?

In this 100-point model, 75 or higher is healthy, 50 to 74 needs attention, and below 50 is at risk. Confirm those lines with your own data by checking what scores renewing clients held 30 to 60 days before renewal.

Next Steps: Build Your Client Health Score This Week

Choose four to six inputs and assign weights totaling 100, with results carrying at least 30.
Score every active client in one sitting in a spreadsheet with a date column.
Write the red, yellow, and green playbooks on one page, with owners and deadlines.
Book a 30-minute weekly review and a quarterly validation date.

The first version will be imperfect. A simple score you review every week will predict renewals better than a perfect model you never build.