Receptive has been acquired by Pendo.

Learn more

How to Create a Health Score that Works Every Time

One of the biggest struggles that customer success teams face is that their health scores don’t accurately represent the ‘health’ of a customer.

One client might show red and renew, while another might show green and churn. This is obviously a huge problem. If left unchecked, your health score is worthless, and you’ll struggle to identify the customers who need your help and attention.

Our Director of CS, Aly Mahan, recently chatted with Irit Eizips for her podcast. Irit is the CEO of CSM Practice and she shared her incredible method of creating a health score that works every time. Every. Single. Time.

Before we go any further, you may wish to listen to the interview. You can do that using the link below.

Listen to Aly speak with Irit Eizips about health scores.

health score

What’s Your Objective?

One of the main lessons that Irit shared was that you should have a strong and clear objective for your health score. What exactly do you want it to measure and predict?

A recent client approached Irit due to their ongoing issues with their health score. The main issue was that the large majority of their customers were yellow according to a new health score which they’d created. This didn’t tell them anything valuable or help them identify the customers most likely to churn.

After their consultation, it became clear to Irit that their problem was not that they hadn’t set the correct objective. There simply wasn’t enough data being collected to accurately predict churn. They restated their objective and decided to use their health score as an ‘early warning system’ as opposed to a ‘churn forecast indicator’. With this new objective in place, yellow meant that attention was required to push them up into green.

By making their objective clear and relevant to their needs, Irit’s client was able to change the way that the CS team operated, enabling them to be more proactive and strategic in their approach.

Signs That Your Health Score is Wrong

During the podcast, Irit shared a couple of warning signs to look out for, that may suggest your health score isn’t going to work.

Too Many Metrics Spoil the Score

Another one of Irit’s clients had eight different metrics that they used to calculate the overall health score. Each of those metrics was given a weight ranging from 5 to 30 percent. The problem here, as Irit pointed out, was that mathematically you would need to have at least four out of the eight metrics to turn red simultaneously to show a customer as being at risk.

This led to a lot of false negatives - customers who seemed fine but in reality were struggling. As far as risk indicators go, it wasn’t doing its job. By the time a customer was flagged as red it was too late to do anything about it.

The solution to this problem, should you find yourself faced with it, is to limit the number of metrics that feed into the overall calculation of your health score. That way a change in one metric will lead to a significant change in the overall score and you’ll notice it right away.

When will you Learn? You can’t Predict Churn

Essentially, any health score that is designed to accurately predict churn is destined to fail.

The other warning sign that Irit spoke about sounds a little counter-intuitive. Essentially, any health score that is designed to accurately predict churn is destined to fail. Irit went to great lengths to hammer that point home because it flies in the face of most health scores.

The reason that trying to predict churn is a recipe for disaster is actually fairly simple. Statistically, most predictive models will only result in 80% accuracy. Even the best statisticians won’t predict churn with 100% accuracy. While 80% accuracy sounds fairly high, it also means that 20% of the customers will have a false positive or negative. And in reality the accuracy will often be even lower due to other factors (such as the aforementioned too many metrics) that come into play. That’s potentially a lot of customers that are labelled incorrectly.

So, how do you avoid those pitfalls and create a health score that actually works every time?

perfect health score

Designing the Perfect Health Score

Okay, we’ve got to the exciting part. I’m going to walk you through the steps as laid out by Irit. If you follow these steps you’ll end up with a health score that works 100% of the time, and allows your CS team to be proactive and strategic.

Step 1: Analyze Your Top Performers

Who are your top customers? These are the ones who are most engaged with your product, your advocates, your most loyal users.

The first step is to identify those top performers. These are your ideal customers and are the users that you want your other users to be.

Once you’ve identified a few of them, try to ascertain what attributes these top performers have. In this case, ‘attributes’ refers to engagement with your product.

Some examples of attributes include:

  • No. of features used

  • Utilization %

  • Advocacy / Advisory Board participation

You get the idea. These are measurements of how engaged and how loyal a customer is to your product and organization.

Step 2: Using the Attributes

Those attributes should be common across all of your top performers. In other words, these attributes determine whether a customer is successful or not.

The attributes can now be used as metrics to help you calculate a health score. Each of these will show you how close a customer is to being your ideal customer, your top performer.

Step 3: Embrace the new Objective

A low score doesn’t necessarily mean a customer is going to churn. Instead, it tells you that the customer needs help and support in order to strengthen their relationship with your product.

You’ve probably worked it out from the previous step, but your new health score also brings with it a new objective.

Rather than focusing on predicting churn (which we’ve already established is a big no-no) your new objective is to help customers become aligned with your top performers.

The healthier the customer, the closer they are to being your ideal customer. A low score doesn’t necessarily mean a customer is going to churn. Instead, it tells you that the customer needs help and support in order to strengthen their relationship with your product.

Using your new Health Score

This new health score flips the old way of doing things upside down. You’re no longer second-guessing whether a customer will churn. You’re now seeing which customers need help, and the areas that you need to focus on. This does two things.

Firstly, it removes any false positives or negatives. Your score is showing you how close a customer is to the top performers. It isn’t predictive, so it’s 100% accurate.

Secondly, your CS team can be proactive in their approach. Rather than waiting for a customer to drop into the red, and then rushing to avoid churn, you can identify the customers who need educating and supporting, and you can move in to help push them to green before it’s too late.

The ultimate aim is to help each and every customer become a best of breed. Doesn’t that sound a lot better than worrying about churn?

If you haven’t already, you should listen to the podcast with Irit. She has a lot more advice to give on health scores that simply couldn’t fit in this article.

Listen to Irit talk about health scores here.