Health scores present a strange challenge for customer success professionals. 

Everyone agrees that they are absolutely critical to understanding the pulse of an account, and that health scores are vital.. But, despite all the emphasis placed on them, very few CS leaders actually trust their health scores.

According to our research, only 3% of CS leaders completely trust their current health scores. 

“If it’s not broken, don’t fix it…but it’s probably broken,” quips Sydney Strader, Catalyst’s VPof CS. 

For the past decade, CS leaders and Customer Success Platform (CSP) providers alike have attempted to do just that. They’ve tried to develop useful health scores using their own formulas and frameworks; and while some do indeed work for specific businesses, none so far can be considered an “industry standard” because of all the variables involved.  

Even our team at Catalyst built an early version of a “standard health score,” but we quickly realized that we needed a completely different approach: weighted health scores (with a twist). 

Weighted health scores are a flexible and data-driven way to get a reliable customer health score. The process of building them involves taking a comprehensive collection of metrics and assigning them priorities or “weights” depending on how your business and customers operate. This strategy makes your customer health scores more accurate so that you can use them to better plan your account strategy,anticipate churn risk, and accurately forecast revenue. 

But does that mean you need to use Catalyst to get a health score you can trust? Certainly not! (Although we WOULD welcome you with open arms.)

In the webinar Untrustworthy Health Scores: What’s the Solution, our aforementioned VP of CS, Sydney Strader, our Senior Product Manager, Susan Tran, and Goldcast’s Director of CS, Sy Pendergast, discussed the key strategies we employed to build our own highly trustworthy customer health score - and how you can build one, too.

Evaluate every aspect of your current score 

Our first health score was a feat of engineering and data, but that didn’t mean it was what we needed.

“If we’re gonna be totally honest,” admits Susan Tran, “the reason we decided to change our first health score was our customers. They weren’t using our scores. They were going out of their way to circumvent what we built in order to build in what they needed.”

It wasn’t enough though to simply know that these scores weren't working; we needed to know why they weren't working. We needed to figure out what it was about the scores that was leading to inaccuracies.

This involved testing every part of the formula, using different variables, and assessing whether the results were actually useful to customers. After all, Catalyst wasn’t going to be the only company using this new health score. 

In your case, it involves figuring out where your current score starts breaking down. What metric throws off the scoring model? Is it a single metric, or a combination of them? Is the way the metrics are being calculated compounding the error?

The more you know about how it’s broken, the better idea you’ll have of how to fix it. 

Identify the most relevant health signals

This step sounds deceptively simple, but it’s where most businesses stumble. 

“Conceptualizing data is extremely hard,” says Susan Tran. “It’s not easy to figure out by yourself what your health score should be and how you should weigh your inputs.” 

Catalyst had to come together as a team to determine how the weighted health function should work, and what specific factors our users should be able to bring in to be part of the formula. It required the cooperation of multiple team members from both the engineering and customer success departments. Other companies were even hiring analysts to help build their own health scores. 

When you are building your own health score, you have to take a retroactive look at your customers and do a comparative analysis based on different segments. 

Which customers renewed, and which churned? What are the commonalities and underlying indicators that you’re seeing as consistent trends amongst these accounts? How big of a factor is product adoption? Or the departure of the executive sponsor? Or the speed of the onboarding? 

Determine which attributes best indicate a customer’s level of satisfaction and willingness to renew, and use that as a foundation for the inputs into your score. 

Pay close attention to that phrase: “Best indicate.” 

“All data inputs are NOT created equal.” - Gloria Liou, Senior CSM, Catalyst

As you construct your score, watch out for metrics that seem very valuable, but are actually the opposite. 

“Customer sentiment is a very controversial measurement,” Susan Tran says. “It depends heavily on the bias of the observer.” Three different CSMs can be on the same call, and each could give a very different impression of the client afterwards. 

And then you have the outliers. These are very specific events  that happen one time, but make such an impact that they get included as a health score factor - even when they shouldn’t be. For instance, your customer may only use your product seasonally, and so usage will be low for most of the time, but the customer is actually healthy. Watch out for these outliers, because they may end up skewing your data if they slip in and get defined as a “danger sign.”

Highlight insights that drive action

The best customer health scores are the ones that help CSMs figure out what to do next to improve the account. 

Inputs that drive action are the most appropriate ones to include in a health score metric. “Why would you include inputs that you can’t actually work on?” Sydney Strader asks. 

This is something we tried to do in our own weighted health scores. “My favorite component of our weighted health is that it’s prescriptive,” Susan Tran says. “It gives direction and indicates to the CSM that if you double down on this one area, it will have the biggest impact on driving health for your customers.”

When you try to construct your own health scores, consider what your CSMs are doing right now to drive action. For example, if an account is currently undergoing an executive escalation, then you can consider that an at-risk indicator. If they’re currently having a case study conversation, then that is a healthy or positive indicator. 

You need to backtrack and determine what inputs drove you to have these conversations with the customer. Those inputs will be incorporated into the health score calculation so that you can more accurately predict the next time those conversations have to take place. 

You can also think of these inputs as milestones for your CSMs. They are Key Performance Indicators (KPIs) for how close a customer is to getting value and achieving their desired outcome(s). 

Build a scoring framework that scales

When you create a health score, don’t assume that it will be the final version. After all, your customers are not static, and neither are your products, your business, or the practice of customer success for that matter.

“Make sure you’re building a system that is flexible,” Sydney Strader recommends. “It has to be able to grow alongside you as you become more knowledgeable and continue to learn every step of the way. 

In fact, it makes sense to start small. Create as lean and as simple a health score as possible at first and build up from there as needs arise. 

Sy Pendergast from Goldcast echoes this sentiment. “Don’t start with 100 different things. It’s going to muddy the waters around what matters and what you can impact. That helped me, because I started with the core things I knew would impact the customer. I was able to add over time, and subtract what was unnecessary. Your health score will change with what’s important to your product and what’s important to your customers.”

That’s one of the reasons we at Catalyst decided to build a weighted health score system in the first place: to create a living, breathing customer scoring model that would grow and change with your business and your understanding of your customers. 

In conclusion

If there’s only one thing you can take away from this article, it’s this: one-size-fits-all health scores do not work. 

Only trust scores that are built from the ground up, based on your business model, and use your standards for what constitutes risk and health.

Home-grown scores are often the most trustworthy due to the simple fact that you know your business best, and the more care and effort you invest into crafting your health score, the more reliable it will be and the better decisions you will make.  

‍Interested in learning more about our Weighted Health Score system? Watch the full webinar Untrustworthy Health Scores: What’s the Solution on-demand.‍