How to Create a Customer Health Score

4 min read

There are many aspects of a customer relationship. There may be an opportunity to sell more, the customer may or may not be happy enough to promote you, and there is always some risk of losing a customer.

While these different aspects tend to be related, it's not always so. For example, you may have a fully adopted customer that's really happy, but currently there is nothing more you can sell them. So despite them being happy with your service, there is no growth opportunity on that account.

The reverse may also be true, with a customer that's only "fairly happy" with your service, and yet may be considering buying more. To avoid any confusion most companies use a Health Score as an indicator of churn risk.

The customer health score is your leading indicator - it is there to help guide you on how and where you should prioritize your time. When designing a health score, you should end up with something basic as:

definitions of red yellow and green score cards for customer health scores

Super basic example:

You design a health score that goes from a scale of 1-10 using only two metrics:

two example metrics to include in your health score

For one of your customers, you find out that:

  1. Latest NPS score given (score measured between 1-10)

  2. Number of logins has been 8 last 10 days (they login frequently) and 2. Latest NPS score given was a 10

You decide to weight number of logins and the NPS score equally and end up with a score of 9 (8+10/2).

9 is a strong number = strong health. Low churn risk!

Designing a Customer Health Score

Designing the actual health score can be incredibly nerdy or super simple (like the example above). Our experience is that a lot of companies struggle with creating their health scores despite all the resources they put into building them.

The reason most companies struggle with creating their health scores is that a health score cannot be both reliable and actionable. Before designing a customer health score, you need to decide the purpose of your health score, do you want it to be a:

creating an actionable health score or a revenue driven health score

The answer to that question will define how you go about building your health score.


If you decide to design a health score that is reliable and predictable for future revenues/churn, then things like “industry” or “contract size” or “time they’ve been a customer” may turn out to be important factors to make it reliable and accurate. However, none of these factors are particularly actionable - it’s hard for your CSM team to affect them. On the other hand, if you decide to design a health score that is actionable for your CSMs, you might see things like “low usage” or “last contacted” to be important (and actionable) factors.

customer health score illustration

After deciding on your approach, you either manually setup the factors for your health score or you take a data driven approach where you do data modelling (or use a machine learning) to help you decide the factors.

In the table below we will try to simplify the pro’s and con’s of both to help you guide you in your approach of setting up a health score.

chart for machine learning health score or manually set up health score pros and cons

Our recommendation

Here is what we recommend

1. Keep it simple

Start with something simple. Most companies/ products health score is mainly a function of product usage (In Planhat we have a concept called the customer “beat”) and your relationship with the customer (basically when you were last in touch with them).

2. Let it take time - improve and adjust

Your health score algorithm might need to change depending on how your product and organization develops. You might want to have different scores for different segments of your customer base. We recommend you to start with something simple and adjust and improve as you learn more and grow your business.

3. Different scales showing churn prediction and up-sale opportunity

Do not include “up-sale opportunity” in the same scale as your churn prediction in your health score. It will complicate things. If you want to assess the likelihood of up-sales, then create another, separate, health score for up-sale opportunities - do not try to show them both in the same scale. Here is a list of metrics your customer health score could incorporate.

examples of metrics to include in your customer health score

Good article on pro’s and con’s of NPS

Read this pdf if you’re new to NPS and want to learn about some of the providers

Join our newsletter!

Receive the latest news, updates, and invitations to our events.

Being data-driven, are we there yet?

Being data-driven should be standard nowadays, but many organizations still struggle with it. Every company wants to be data-driven, but putting it into practice is the tough part.

How to transition from a cost center to a profit center in CS

Many in CS shy away from the commercial aspects of the business, but it's a missed opportunity to keep the conversation moving around maximizing and driving value for customers.

It's worth the risk: Identifying and managing risk in CS

When faced with risk in CS, it’s hard to know what the first step should be. We've developed an actionable plan based on experience from three CS experts.

Learn more about

Drop your email and let us show you our platform!