K
KnowMBAAdvisory
RetentionAdvanced7 min read

Customer Health Score

A Customer Health Score is a composite metric (typically 0-100) that predicts whether a customer will renew, expand, or churn. It combines product usage data (login frequency, feature adoption), engagement signals (support tickets, NPS responses), and business outcomes (ROI achieved, time-to-value). Gainsight data shows that accounts scoring above 80 renew at 96%, while accounts below 40 churn at 55%. Proactively reaching out to at-risk accounts can save 20-30% of them.

Also known asHealth ScoreCustomer HealthAccount Health ScoreUser Health IndexCHS

The Trap

The trap is building a health score based on vanity metrics like 'total logins' instead of value-delivered metrics like 'completed workflows.' A customer logging in daily to export data to a spreadsheet (because your reporting is broken) looks 'healthy' by login frequency but is actively seeking alternatives. Similarly, weighting NPS too heavily ignores that a promoter (NPS 9) with declining usage is more at-risk than a passive (NPS 7) with increasing usage.

What to Do

Build a weighted health score with 3 categories: (1) Product Engagement (40% weight): DAU/MAU ratio, core feature usage, breadth of feature adoption. (2) Relationship Signals (30%): NPS/CSAT trend, support ticket sentiment, executive sponsor engagement. (3) Business Outcomes (30%): ROI achieved vs promised, time-to-value, expansion usage. Score each 0-100, weight and combine. Set alerts: Green (70+), Yellow (40-69), Red (0-39). Review all Red accounts weekly with a playbook for intervention.

Formula

Health Score = (Product Engagement ร— 0.4) + (Relationship Signals ร— 0.3) + (Business Outcomes ร— 0.3)

In Practice

HubSpot built their legendary Customer Health index by defining 'CHI' (Customer Happiness Index). Instead of just looking at logins, they tracked whether customers were adopting the features that made them successful (like setting up a landing page or launching a campaign within 30 days). A high CHI score meant a customer was achieving their marketing goals; a low score meant they were stuck. HubSpot used this score to proactively reach out to struggling customers, drastically reducing their early churn rate.

Pro Tips

  • 01

    Recalibrate your health score quarterly by comparing predicted vs actual churn. If your model says 85/100 accounts are healthy but 15% still churn, your weights are wrong.

  • 02

    Track health score velocity (trend), not just the absolute number. An account dropping from 85 to 65 in 30 days is more at-risk than one stable at 50 for 6 months.

  • 03

    Add 'negative signals' with outsized weight: competitor mentions in support tickets, admin password resets (someone new is evaluating), CSV exports of all data (migration prep).

Myth vs Reality

Myth

โ€œHigh usage always means a healthy accountโ€

Reality

Usage without outcome is a red flag. A team using your project management tool 8 hours/day but consistently missing deadlines isn't getting value โ€” they're just stuck. Measure outcomes (projects completed on time), not activity (hours spent in app).

Myth

โ€œYou need machine learning to build an effective health scoreโ€

Reality

Start with a simple weighted average of 5-8 signals. Gainsight, ChurnZero, and Vitally all started with manual scoring rules. ML adds 10-15% accuracy improvement but requires 10,000+ data points. Simple rules get you 80% of the way with 1% of the effort.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge โ€” answer the challenge or try the live scenario.

๐Ÿงช

Knowledge Check

Account ABC has a health score of 82/100. However, their DAU/MAU ratio dropped from 45% to 22% over the last 60 days, while their NPS score remained at 8/10. What should you do?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets โ€” not absolutes.

Health Score Distribution (% Green)

B2B SaaS, mid-market accounts ($10K-100K ACV)

Elite

> 75% Green

Good

60-75% Green

Average

45-60% Green

Needs Work

30-45% Green

Critical

< 30% Green

Source: Gainsight 2024 Customer Success Benchmarks

Intervention Save Rate

At-risk account recovery (Red zone accounts)

Elite

> 35%

Good

25-35%

Average

15-25%

Needs Work

8-15%

Critical

< 8%

Source: ChurnZero 2024 CS Industry Report

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

โš™๏ธ

PTC

2018-2021

success

As PTC transitioned from selling perpetual licenses to SaaS subscriptions, they realized their historical metricโ€”support ticket volumeโ€”was useless for predicting churn. They implemented a sophisticated health score tracking feature adoption breadth. They discovered that customers using 3 or more modules renewed at 98%, while those using only 1 module renewed at 65%.

Predictive Signal

Module adoption breadth

High Risk

1 module (65% renewal)

Healthy

3+ modules (98% renewal)

Feature adoption breadth is often the strongest leading indicator of health in a complex B2B product. If an account relies on multiple parts of your platform, their switching costs are high and they are extracting deep value.

๐Ÿข

WeWork (Enterprise)

2017-2019

failure

WeWork's enterprise division tracked 'health' by the physical utilization rate of desks. If an office was 80% occupied, they considered the account healthy. However, they ignored relationship signals and business outcomes. When a major tech company's lease came up, WeWork assumed they would renew because desk utilization was high. Instead, the company churned because their internal employee satisfaction with the spaces was terrible, and the decision-maker had changed 6 months prior.

Usage Metric

80% desk utilization (Green)

Relationship Signal

Loss of champion (Ignored)

Outcome

Unexpected churn of $5M account

Usage does not equal health. If you track engagement but ignore relationship signals (like a change in the executive sponsor) and outcome metrics, you will be blindsided by churn from 'highly engaged' accounts.

Decision scenario

The Watermelon Account

You review the health dashboard for your largest enterprise client ($500K ARR). The overall health score is 75 ('Green'). Your renewal is in 90 days. But when you look closer, the score is 'watermelon': green on the outside, red on the inside.

Overall Score

75/100 (Green)

License Utilization

95% (Perfect)

Support Sentiment

30/100 (Red)

01

Decision 1

While 95% of the licenses are active daily, the support team is drowning in aggressive tickets from this account complaining about system crashes. The engineering team says the fixes will take 6 months. The account's executive sponsor has gone 'dark' and isn't replying to your emails.

Trust the overall score. Usage is the ultimate truth. Send them the standard 90-day renewal contract.Reveal
The executive sponsor finally replies: 'We are cancelling. Our team is using the product because we force them to, but it crashes constantly. We bought a competitor last week.' Being blindsided by a high-usage account is the classic health score trap.
ARR Lost: $500KHealth Score: Failed to predict churn
Declare 'Code Red'. Override the overall score due to the support sentiment and dark executive. Escalate to your CEO to fly out for an in-person meeting.Reveal
Your CEO flies out and discovers the executive sponsor left, and the new leader hates the crashes. Your CEO presents the 6-month product roadmap directly, apologizes, and offers a 3-month SLA credit. The client agrees to a 1-year renewal strictly based on this executive intervention.
Intervention: Executive alignment restoredARR Saved: $500K

Related concepts

Keep connecting.

The concepts that orbit this one โ€” each one sharpens the others.

Beyond the concept

Turn Customer Health Score into a live operating decision.

Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.

Typical response time: 24h ยท No retainer required

Turn Customer Health Score into a live operating decision.

Use Customer Health Score as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.