How to create a customer health score with four metrics
If you want to understand customer health scores, think about your last doctor visit.
While you were there, a nurse probably measured your body temperature, pulse rate, and respiration rate. Those vital signs didn't make an exhaustive diagnostic but they gave your doctor a general idea of your health. Even more, they acted as new data points that could be compared to previous vitals, making it easy for your doctor to find dangerous trends.
A customer health score is similar. It’s an index made up of several “vital” key performance indicators (KPIs). It gives you an overall idea of how healthy a customer’s relationship is with your product. It also acts as a fortune-teller—predicting future customer behaviors like renewing or canceling subscriptions.
Customer health scores can get pretty complicated. We’ve seen some that include dozens of metrics. But it doesn’t have to be that way. You can build a useful health score from a handful of inputs. After all, your health score won’t be helpful if no one understands it.
Let’s do exactly that. We’ll use a simple three-step process to build a customer health score based on just four customer health “vital signs.”
Step 1: Choose the metrics for your customer health score
For software developers, a minimum viable product (MVP) is the least complicated version of a product that can still provide value to its users. We’re going to hijack that method and build an MVP health score using four metrics:
Product usage rate
Net promoter score
We looked at the data from hundreds of customer health scores and realized these KPIs occur commonly. Most likely, it's because they offer a mix of qualitative and quantitative inputs, which gives you a more holistic view of the customer.
Product setup (aka product adoption or product implementation) tracks the steps your customer needs to take to get full value from your product.
There are two ways to look at product setup:
A total number, i.e. how many implementation steps a customer has taken thus far
A percentage of the total steps available, i.e. how far a user has progressed in their adoption journey
At Vitally, we ask users to create Playbooks, set risk alarms, and integrate with other software. If a user does all these things, we know they’re getting value out of our product. That's why product setup is a great metric to have in our customer health score.
Product usage rate
Product usage rate measures how often users engage with your product in some meaningful way over a set period of time.
When calculating your product usage rate, you have to do two things:
1. Make sure your usage rate isn’t just a measure of someone opening the app
For example, in a collaboration and productivity app, users should be creating and completing tasks regularly. If they’re just opening the app and not managing tasks, that’s probably a sign of trouble.
2. Normalize the usage rate for population
Population is the number of people using your app (aka users, seats, or licenses) at a given customer account. Normalizing for population means averaging the number of times your product gets used by the number of users.
Asana has customers that range from a lone consultant to a massive team like Salesforce. If the lone consultant created and completed 10 tasks this week, that’s a great outcome. But if the entire Salesforce team created and completed only 10 tasks this week, that’s bad. If you look at the total times a customer account used your product, but you don’t adjust for the number of users in that account, the results don’t tell the full story.
To normalize for usage rate, divide the total uses by the number of users.
Let’s say Salesforce had 100 Asana users in total. Those users created or completed 500 tasks in a day. Their daily product usage rate is 5.
Product usage rate is an important metric because if someone isn’t using your product, they’re not very likely to keep paying for it.
Net Promoter Score
A net promoter score (NPS) is a customer satisfaction benchmark. It’s based on a single survey question that gets sent to your customers: “On a scale of 0 to 10, how likely are you to recommend our product to a friend or coworker?”
Calculating Net Promoter Score
Responses to that question are typically placed into three groups:
1 - 6 = “Detractors”
7 - 8 = “Passives”
9 - 10 = “Promoters”
To calculate NPS, subtract the percentage of detractors from the percentage of promoters (ignoring the passives).
Let’s say Asana sends an NPS survey to 100 users at Salesforce. Of that group, 70% were promoters, 20% were passives, and 10% were detractors. That customer account has an NPS score of 60%.
People don’t recommend what they don’t like, so NPS scores are considered a reliable proxy for customer satisfaction and loyalty.
The CSM pulse is a subjective grade that your Customer Success Managers (CSMs) give each account they manage. Think of it as a human “gut check” on the happiness of your customers.
You can create just about any scale you want for the CSM pulse. To make things easy, we’ll use a 0 to 10 scale for this exercise.
You can—and eventually should—measure a whole slew of customer engagement metrics like the frequency of CSM-to-customer contacts and the number of service tickets reported. For the first customer health score, the CSM pulse is a simple way to express customer satisfaction in one tidy number.
Step 2: Assign a common scale to each customer health score metric
Our ultimate goal is to calculate a single health score that reveals the overall health of a customer. One metric to rule them all, as it were.
To do that, we have to create a common scale for each of the metrics that make up the score. If we don’t, the calculation would be impossible—how do you reconcile an NPS of 60 with a product usage rate of 5?
It’s pretty easy to do. Our scale is going to consist of three categories, and each gets 0, 5, or 10 points:
Poor (0 points)
Concerning (5 points)
Healthy (10 points)
Then we’ll decide what poor, concerning, and healthy means for each metric. There’s no hard and fast rule here—a product usage rate of 5 might be healthy for your business but poor for another.
We created this matrix for our example:
Let’s run through this so you can see how it works
Your customer, Salesforce, has these results:
Product setup rate: 80% - that’s 10 points
Product usage rate: 6 - that’s 5 points
NPS score: 32 – that’s 5 points
CSM Pulse: 9 - that’s 10 points
You now have what you need to calculate a simple customer health score for Salesforce. Just add up all the points and divide by four (the number of metrics we’ve used).
You could stop here if you wanted to, but you probably shouldn’t. Why not? Because some of the metrics we’ve chosen might be better predictors of churn than others—which means you’ll want to give them more weight in the overall customer health score calculation.
Step 3: Weighting your customer health score metrics
Weighting the metrics in your customer health score will make that score a much better predictor of customer behavior.
What if you noticed that 85% of your churned customers had a product usage rate of less than 5? That’s a strong correlation. Meanwhile, you’ve noted that only 40% of churned customers had a low CSM pulse score—a much weaker correlation. In this case, product usage rate is a better predictor of customer churn.
So, we’re going to give product usage rate more weight in our customer health score.
Weights (need to add up to 100)
Product setup rate: 20%
Product usage rate: 40%
NPS score: 20%
CSM pulse: 20%
Multiply the raw scores by the weighting factors to get our weighted net scores.
Did you notice what happened? When we weighted the metrics, Salesforce’s health score dropped from 7.5 to 7. They didn’t score well in the most predictive category we tracked: product usage rate. If you hadn’t weighted the score, you might have missed an important churn signal.
A final note: Your customers' current health score is only a small part of the story. The real insight is in the trend over time. Salesforce might have a health score of 7 right now, and that’s not too bad. But if it was 8 last month, there’s cause for concern.
Customer health score: What’s next?
Once a minimally viable software product is launched, developers typically go back and start iterating on it. That’s exactly what you can do with your customer health score.
You can start by reviewing some of the dozens of other customer and product metrics out there—at Vitally we like data around support tickets, customer engagement with the company, and account growth (like upsells). What you’re looking for are those crystal ball analytics that predict customer behavior.
You can also look for ways to create new customer health scores for specific customer segments. That way, you’ll score new customers differently than established product users.
But for now, take a breather. You have an effective health score using just four customer "vital signs." You can start using that score today to find those hidden churn risks and low-key upsell potentials.
The right tech tools will automate repetitive tasks, expose insights from large quantities of data, and help disparate teams collaborate more effectively. But with hundreds of software solutions available, how do you choose? Here's a list of 11 tools
Customer churn prediction is hard—we get it. Especially when you consider most customers don't tell you when they're unhappy. Here are 5 mistakes you might be making when predicting customer churn and what to do about them