How to build a customer health score from scratch

While anecdotal data from conversations is useful in determining a customer’s health, many teams use a customer health score to better understand the status of their customer base. These scores help customer success (CS) teams (and the broader organization) take a more proactive approach, rather than only reacting to inbound requests and complaints.

In a recent episode of the Churn It Up podcast, we spoke with Rachel Jennings, a customer success manager (CSM) at Assignar. Rachel explained how she developed a health scoring system to help her CS team better understand which customers were healthy, which were at risk of churning, and which might need more intervention.

Whether you already have some sort of health score in place, you know you need a customer health score but don’t know where to start, or are starting completely from scratch, here are the steps to building a customer health score that your entire organization can benefit from.

Step 1: Ask the right questions

A key first step in building a customer health score is understanding the motivation behind your efforts, and what this score should represent. Here are four questions Rachel asked herself in the early stages of developing her health score:

1. 1. Do we know the health of our customers? If you’re creating a health score for the first time, the answer to this is likely no. For the Assignar team, they had anecdotal evidence of customer health, but they were lacking a data-driven answer to this question.
   2. If we put a health score in place, what does that actually mean to us? It’s important to identify a clear reason for using customer health scores. Rachel said they planned to use the score to help her team assess whether a customer is at-risk or healthy and moving toward their “Assignar Champion” status.
   3. What do we define as a healthy customer? This will be different for every company, but Assignar determined that a healthy customer used the product at least five days a week. The team also identified key habits in the product to look for and habits that signified an engaged customer, for example giving feedback and providing testimonials.
   4. What do we define as an at-risk customer? There are a lot of variables to look for: a reduction in usage, too many (or no) support conversations, low NPS scores, no longer giving references or participating in events, or payment or billing issues. An at-risk customer is likely some combination of all of these factors (and more), but it depends on your business and your product.
      E-BOOKLearn what ten experts believe it takes to be a successful data-driven product manager

Step 2: Decide what you want to measure

Once you’ve thought through these questions, you need to decide which metrics will actually contribute to your health score. This can be difficult since there are so many different metrics you can track, but Rachel recommended keeping it to five to ten. One useful tactic is to list out everything you’d like to measure, and then you’ll start to notice some themes–making it easier to group metrics together and choose the one that will be most representative.

The Assignar team broke their metrics out into two categories: product metrics and relationship metrics. Here’s what goes into their customer health score:

プロダクト：

• • Bugs
  • Issues users experience
  • プロダクトの定着率
  • Product usage (e.g. daily active users, monthly active users, or stickiness)
  • Interactions with support

Relationship:

• • NPS
  • Engagement (referring other customers, providing testimonials, submitting reviews, participating in marketing activities, etc.)
  • Expansion opportunities
  • Financial (e.g. payment amount or whether or not they pay on time)

Pro tip: You can use your customer health score to examine health at the user level and the account level.

Step 3: Determine the weight of each component

Not every part of your customer health score will have the same weight. Once the metrics were finalized, Rachel then determined how she would weigh each metric. For Assignar, it made sense to weigh product metrics at 60 percent and relationship metrics at 40 percent (with further breakdowns in each category). She then created a Health Score Matrix to map out the weight of each metric, where each is scored between one and five.

To provide an easy understanding of their customer health score, Rachel decided to break the final score into five categories ranging from the customer is at-risk/going to churn to the customer is a champion. Again, the exact breakdown and meaning of your score will be different for every company, but this method is a good framework to follow.

Step 4: Automate the process

Rachel suggested first setting up customer health score calculations manually (to ensure the system works), and then eventually exploring ways to automate the process. If you already have a product analytics tool, you can explore ways to automatically push some of this data into your scoring model, and also segment product usage data by score to understand what behaviors drive healthy vs. at-risk customers.

The benefits of customer health scores

The most obvious benefit of a customer health score is that it enables the CS organization to have a much more process-based approach to their work. Rather than getting pulled in many different directions (usually by the loudest customers), a customer success team with health scores in their arsenal can take a more proactive approach, as they are able to see which customers need help based on data.

The benefits also go beyond the customer success org. Having a clear system in place makes it much easier to share this information with the rest of the company, especially if you can create shareable dashboards that include an account’s health score. This also makes it easy to report back to the leadership team on metrics like the amount of revenue at risk of churning for a given quarter.

If you want to listen to the full podcast episode, check it out here. You can also read more about how Rachel developed her customer health score matrix here.