Cracking the code to customer health scores

No business can thrive without a clear understanding of its customers' health. Customer health scores act as the early warning system for retention and growth, helping businesses identify risks, maximize expansion opportunities, and drive long-term success.

From tracking usage trends and engagement levels to measuring value realization, customer health scores offer a data-driven approach to understanding customer relationships.

But while many companies attempt to implement them, few get them right – leading to misleading insights, missed churn signals, and ineffective interventions.

So, what makes a customer health score truly effective?

In this guide, we’ll break down:

Understanding customer health scores
Why traditional customer health scores fail
How to build an accurate and actionable health score
Best practices for taking action on health insights
The importance of continuously refining your scoring model

Understanding customer health scores and why they fail

Customer health scores are a key metric used to assess the likelihood of a customer churning, renewing, or expanding their relationship with your business.

The most accurate ones blend both quantitative and qualitative data, allowing organizations to proactively mitigate churn risk and identify growth opportunities.

These scores are typically presented as a numerical value, often categorized into red, yellow, and green – terminology that many customer success teams are already familiar with.

Why traditional customer health scores fail

Despite their potential, many organizations struggle to make customer health scores effective. Here are the key reasons why they often fall short:

1. No alignment with the customer journey

One of the biggest issues is the lack of alignment with the customer journey. If your organization hasn’t clearly defined the customer lifecycle – from onboarding and adoption to advocacy – it becomes difficult to take meaningful action when a score indicates risk.

The first step should be assembling a cross-functional team to map out who is responsible for each stage of the customer journey. Without this foundation, even the best health scoring models will be ineffective because you won’t know what proactive steps should have been taken in the first place.

2. Data silos prevent a holistic view

Customer data often resides in multiple systems, making it difficult to get a comprehensive view of customer health. If your data isn’t unified into a single, accessible view, your customer health scores will lack accuracy and usability.

Instead of driving proactive action, they become just another static report, leaving teams in a constant reactive state.

3. Over-reliance on manual processes

If you’re building customer health scores in a spreadsheet, you’re already missing out on critical opportunities to mitigate churn and drive expansion.

Manual processes are slow, prone to errors, and don’t allow for real-time insights. Automation is key to ensuring your scores are up-to-date and actionable.

4. Too much focus on either qualitative or quantitative inputs

Another common pitfall is relying too heavily on either qualitative or quantitative data. Many organizations lean too much on usage metrics, assuming that high engagement equals high retention. This often leads to what I call watermelon customers – green on the outside, red on the inside.

These customers may appear healthy based on product usage but ultimately churn for reasons that weren’t captured in the scoring model.

On the other hand, relying too much on qualitative inputs like surveys or customer sentiment can be equally misleading. Feelings and perceptions, while important, don’t always translate to real business outcomes. Striking the right balance between qualitative and quantitative data is crucial.

5. Misplaced trust in NPS as a key health metric

Many companies use Net Promoter Score (NPS) as a core indicator of customer health. We did too at Brazen – our CFO closely tracked NPS month over month.

But a study conducted by Greg Danes, also known as the Churn Doctor, revealed something surprising: there is no correlation between NPS scores and customer retention. This was a major revelation, as NPS has long been viewed as a gold standard in customer success.

Similarly, customer satisfaction (CSAT) surveys were also analyzed, and the results were even more counterintuitive. The study found that while there was a 12% increase in CSAT, those same companies experienced a 22% decrease in net retention.

In other words, making customers feel satisfied didn’t necessarily keep them from leaving.

Shifting the focus: Why do customers stay?

Rather than asking why customers leave, Greg Danes suggests that organizations should be asking, why do customers stay? This shift in perspective can lead to a more meaningful and predictive approach to customer health scoring.

By identifying the specific factors that contribute to long-term customer retention, companies can build health scores that truly reflect business outcomes, rather than just surface-level engagement or sentiment.

Why do customers stay? The key to meaningful customer health scores

A fundamental question we should be asking is not just why customers leave, but rather, why they stay. The answer to this question holds the key to building truly effective and predictive customer health scores.

The foundation of retention: Value realization

When I ask people why customers stay, the most common responses are relationships and solving a business problem – both of which ultimately tie back to value realization. If there’s one major takeaway about customer health scores, it’s this: Focus on value.

If there’s a single metric or KPI that can predict whether a customer will renew, expand, or churn, it’s whether they are seeing value from your solution. However, identifying and tracking value can be challenging because value is different for every customer.

The problem with traditional metrics

Too often, organizations try to measure value through product usage or customer engagement metrics, such as:

How many QBRs (Quarterly Business Reviews) are we conducting?
How many surveys are we sending?
How many support tickets are customers submitting?

But none of these inherently tell us if a customer is getting value from our solution. They are proxies for engagement, not impact. A customer might attend every QBR, but if those meetings don’t drive meaningful improvements for them, the engagement doesn’t translate to retention.

Value often exists beyond the numbers

Sometimes, the most significant value a customer experiences is intangible – something you wouldn’t necessarily find in traditional metrics.

Let me give you an example.

We worked with a large school district that used Brazen’s virtual and in-person hiring event software to run their career fairs. Their feedback? Our platform played a significant role in narrowing the gap in their first-day teacher vacancies, ensuring they had a high-quality teacher in every classroom on day one.

It also increased leadership capacity across their district and even shifted the mindset within their organization, reinforcing that everyone is responsible for recruiting, not just the talent acquisition team.

None of that insight would have surfaced through product usage data, survey results, or support interactions. Yet, this is the real value they gained from our solution – and understanding this is crucial to effective health scoring.

Building meaningful customer health scores: Where to start

When building customer health scores, the key is to start simple. It’s easy to feel overwhelmed, especially in large organizations with multiple products and customer segments.

When Brazen was acquired by Radency, I went from working with a single-product customer health score to navigating a seven-product solution. The Radency team was unsure where to begin, and like many companies, they were still using spreadsheets and gut feelings to categorize customers into red, yellow, or green based on perceived account relationship strength.

To move forward, we simplified the process by asking a few key questions:

What is the top reason for churn in the last 6–12 months?
Which product should we focus on first? (For Radency, this was Career Sites.)
Are there one or two KPIs that can help track churn risk for this reason?

If you can track value realization, that’s the best place to start. If not, begin with usage metrics – but be mindful of whether customers are churning due to lack of usage or lack of value despite high usage.

In Radency’s case, job volume dips were the strongest indicator of churn for Career Sites, so we started tracking that.

Key elements of an effective health score

1. Segment customers appropriately

Not all customers should be treated the same. A large enterprise customer will have different risk factors than a small business, and different use cases require different scoring approaches. Segmenting your customers ensures that your health scores reflect their unique needs and behaviors.

2. Automate key adjustments

One of the most overlooked aspects of health scoring is building automatic overrides based on real-world factors. For example:

If a customer is actively using your platform but has told their CSM they’re going out of business, their score should automatically turn red – regardless of their engagement.
If a customer’s usage data looks poor but just signed a renewal contract, their score should be adjusted accordingly.

Most CRMs, including Salesforce, allow you to reference these fields and automate adjustments so that health scores reflect reality, not just numbers.

3. You don’t need a data analyst to identify KPIs

A common misconception is that you need a data scientist to define KPIs for your health scores. That’s no longer true.

AI-powered tools can now analyze historical churn and renewal data to surface top predictive KPIs and their weightings.
Experienced customer success reps often already know the leading indicators of churn based on years of working with customers.

At Brazen, we initially had an entry-level data analyst build a complex model using machine learning – but in the end, the most seasoned reps had already identified the key churn predictors without any advanced modeling.

A structured approach to building your customer health score

If you’re unsure where to begin, here’s a step-by-step framework:

Start with usage KPIs

Identify 5–10 usage metrics correlated with retention.
Assign weightings based on historical data (e.g., customers with low usage in these areas are more likely to churn).
Example: At Brazen, we tracked eight usage KPIs and weighted them differently based on their churn correlation.

Incorporate adjustment factors

Reference qualitative inputs from your CRM, CSMs, and renewal data.
Example: If a customer just renewed but has low usage, adjust their score manually to prevent false red flags.

Use a flexible scoring system

Avoid rigid models – your business will evolve, and your scoring model should adapt over time.
At Radency, we used Sigma (a BI tool) because it allowed us to handle complex parent-child account relationships that weren’t easily managed on other platforms.

Ensure your tool integrates with key systems

Your health score should pull data from:

CRM (e.g., Salesforce) – to track renewals and account attributes.
Product usage data – to understand engagement.
Support systems (e.g., Zendesk) – to incorporate ticket trends and pain points.
CS platforms (e.g., ChurnZero, Gainsight) – to enable proactive workflows.

Leveraging AI for predictive analytics

One of the biggest advancements in customer health scoring is AI-driven analytics. Many platforms now dynamically identify:

Which KPIs are most predictive of churn or retention
What weightings should be assigned to different factors
Forecasts for future customer health trends

At Radency, we built some predictive forecasting into our Sigma-based health scoring system, but many customer success platforms now offer this natively. If you’re looking for a tool, consider solutions that offer AI-powered insights and automation.

Automating customer health responses

For every possible change in a customer’s health score, we have predefined automated workflows. Below is a breakdown of how Radency handles different scenarios:

1. A key KPI worsens

Not every critical customer metric is included in the health score, but that doesn’t mean it should be ignored. If a key leading indicator of churn starts trending downward, we take immediate action.

Example:

A customer hasn’t hosted a virtual or in-person event in the last 30 days (a major usage drop for our platform).
We trigger an automated four-email engagement campaign that appears to come from their CSM.
Each email is designed to re-engage the customer and encourage them to take action.
We track the effectiveness of these campaigns month over month – are customers responding? Are they creating events again? If not, we adjust our approach.

2. A customer’s health score is stalled

Many companies focus only on red accounts but stalled accounts in the yellow zone are equally important.

Our approach:

If a customer remains in a yellow health score state for two consecutive months, we take action.
Every week, our team reviews yellow-score accounts to determine if they should be formally flagged as at risk.
If necessary, an auto-alert is sent to inspect the account further.

3. A customer moves from yellow to red

This is a clear escalation and requires immediate attention.

Our process:

The system triggers an auto-alert to our customer success team to inspect the account.
However, not all red scores indicate true risk – for example, some customers have seasonal usage (e.g., they only run two flagship events per year).
A manual review determines if the red status reflects an actual risk.
If the account is confirmed at risk, we mark it in CRM with a designated checkbox.

4. A customer is officially flagged as at risk

Once an account is formally marked at risk, we shift into full retention mode.

What happens next:

An automated email is sent to key internal stakeholders, outlining the renewal date, the reason for risk, and the CSM responsible for handling the account.
A structured at-risk protocol is initiated.

5. The at-risk protocol: Week-by-week playbook

At Radency, we’ve developed a systematic playbook for handling at-risk customers.

Every week, we send email reminders to the responsible CSMs and managers.
Each email includes specific action items aligned with our eight-week at-risk recovery plan.
The goal? Resolve risk factors and re-engage the customer as quickly as possible.

By following this structured approach, we’ve achieved 35–70% retention rates for at-risk customers, depending on how early the risk was identified.

Pro tip: Assign an at-risk operations manager

One of the most impactful changes we made was assigning an at-risk ops manager. This person acts as the central owner of the at-risk process, ensuring that:

The playbook is being followed consistently
At-risk customers are flagged early (no late red flags)
Customers flagged at risk are given proactive attention

The at-risk ops manager can also be measured and goaled on:

Reducing churn among previously flagged accounts
Decreasing the number of accounts that churned without ever being flagged
Minimizing the time it takes to identify and act on risk

How many companies openly share their customer health scores with customers? The answer, surprisingly, is very few – but I strongly believe that sharing at least some components of the health score can lead to more meaningful conversations and better customer success outcomes.

One of the most common concerns I hear is:

“Sales teams don’t want to share it.”
“What if customers react negatively?”
“What if they see a low score and get upset?”

These fears are understandable, but they also prevent us from having open, productive discussions about customer success. If we’ve already analyzed the patterns of customers who churn vs. customers who renew, why wouldn’t we use that insight to guide conversations with customers?

At Radency, we do share a version of our customer health score with customers – but we’ve found that how you share it matters.

It fosters transparency and trust – Customers appreciate knowing where they stand.
It creates a more collaborative partnership – Instead of us assuming their success, we invite them into the process.
It drives more meaningful discussions – Instead of vague check-ins, we have data-driven conversations about their usage and value realization.

However, the key to making this work is what you share and how you frame it.

The version we share with customers is purely usage-based – and we’re very clear that a green score does not necessarily mean they’re seeing value. Instead, it’s a starting point for a conversation:

“Here’s your current usage health score. How does this correspond to the value you’re seeing?”
“Your score is red, and you’re telling us you’re not seeing value – this confirms that we need to increase usage.”
“Your score is green, but you’re still not seeing value – this signals that we need to dig deeper and ensure we’re solving the right problems for you.”

To make this as seamless as possible, we’ve embedded the customer health score directly into our platform using Sigma. Customers can log in and see:

Their real-time usage trends (updated daily).
How their usage compares to other customers of similar size and use case.
A predictive analytics forecast showing whether they are expected to remain in the green zone or decline.

This helps answer one of the most common questions customers ask:

“Is my usage normal?”
“How do I compare to similar companies in my industry?”

By providing this visibility, we empower customers to take ownership of their success while also giving our team an opportunity to guide them toward better outcomes.

Handling industry and privacy concerns

Not every customer will want their data included in aggregate benchmarks. In industries like healthcare and education, data-sharing restrictions may apply.

To accommodate this, we exclude any customers who have opted out of data aggregation in their contracts. For customers who allow it, we provide anonymous, aggregated benchmarks so they can compare themselves against peers.

Despite these considerations, most customers love having access to this data – it gives them insights they wouldn’t otherwise have and helps them track their progress over time.

Continuously refining your customer health scores

Building a customer health score isn’t a one-time project – it’s an ongoing process that requires continuous refinement.

The goal is to ensure that your health score remains an accurate predictor of customer behavior, helping you take proactive action before it’s too late.

At Brazen, we initially identified lack of usage as the primary reason for churn. This made usage metrics the foundation of our health score.

Over time, as we successfully increased customer engagement, we realized that churn was no longer driven by lack of usage – instead, it was due to customers not seeing enough value.

This shift reinforced an important lesson: Your health score should evolve alongside your customers.

The key takeaways

Focus on value, value, value. Usage is a starting point, but true retention comes from ensuring customers realize value.
Start simple, then refine. Begin with foundational metrics and expand as you learn more.
Use a flexible scoring system. Your model should evolve as customer needs change.
Automate as much as possible. Let your CSMs focus on relationship-building while automation handles the alerts.
Share a customer-facing health score. It fosters transparency and drives better conversations.
Continuously refine your model. Set accuracy goals, track performance, and adjust as needed.

Following these best practices ensures that your customer health score remains an effective, dynamic tool – helping you reduce churn, drive expansion, and strengthen customer relationships.