How the wrong kind of customer health scoring can hurt you

If your business relies on renewal revenue (and, with the rise of the subscription-economy, it very likely does), then you are most definitely familiar with customer health scores. They are critical to monitor to ensure that you’re delivering continuous value to your customers and growing your company. In fact, they are so important that many companies with a recurring revenue model are now building customer success teams tasked with ensuring customer retention and growing customer lifetime value.

A customer health score often directly measures the effectiveness of these teams, along with churn and renewal KPIs. But what’s the payback from that score? And how do you determine whether the score truly correlates to customer retention and renewal rates?

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":1777312,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,enterprise,entrepreneur,marketing,","session":"A"}']

Moving from reactive to proactive health scoring

Whether it’s a traffic light model (red, yellow, and green indicators) or a 1-100 scoring metric to identify which customers are happy and which are at risk, too often, these types of customer health scores are based on subjective input and not on leading indicators of customer health. They are reactive versus proactive. They may work, but there is really no way to measure whether they do until it is too late and the customer has already churned. If you want to check if your health score is working before that happens, you need to use predictive analytics techniques.

There are many different predictive models that you can build to detect if your health score is a meaningful representation of customer health. Predictive lift is the key to understanding your health score when you have clean historical data that can be used to characterize behaviors and identify patterns. Predictive lift measures the performance of your health score against random guessing, or what the results would be if you didn’t use a health score at all. By using historical data where you know the outcome of the scenario, you can evaluate how well your health score works to predict certain behavior.

If the predictive lift is high and actionable, the score has good economic value, because you’ll use it to apply resources to priorities that impact revenue retention and growth. But if the predictive lift of a health score is low, its economic value is low, because acting on the score is only slightly better than random or “gut-based” customer interactions.

Information value: the real key to meaningful health scores

In predictive analytics, we have the notion of “information value,” which quantifies how strongly a particular factor contributes to prediction. We find information value through a calculation involving the statistical correlation to an outcome. Information value is normalized to be anywhere from zero (i.e., the same predictive value as random choice) to 1 (i.e., perfect prediction). In practice, a factor that has an information value somewhere above 0.5 will provide good predictive lift and will correlate to decreased churn and improved renewal performance.

Let’s walk through some examples. When looking across its whole customer base, a given company might have 15 percent churn. So from a probability standpoint, if you were to randomly select a customer to approach as a retention risk, there’s a 15 percent chance that you’d actually be engaging an at-risk customer.

Predictive analytics can transform that 15 percent probability into a 50 percent probability or better, and that’s the goal of a health score — to increase the probability of engaging the right customer with the right interaction.

But what if you don’t know the information value of your health score? Suppose it has an information value of just 0.05? Using that health score, the probability that you’re engaging an at-risk customer rises to 21 percent — just a 6 percent lift above not using the health score at all.

A perfect prediction would identify 150 customers — i.e., the 15 percent out of 1,000 that are actually at risk. But with an information value of 0.05, 430 are identified as at risk, and only 90 of those are actually at risk. Consequently, 80 percent of customers flagged as unhealthy would actually be healthy — so you’d be spending time and resources on many customers who aren’t at risk. In addition, 40 of the actual at-risk customers are predicted to be healthy, so you’ll get unanticipated churn.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":1777312,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,enterprise,entrepreneur,marketing,","session":"A"}']

So what happens as the information value increases? At an information value of 0.4, the probability that a predicted churn would actually churn goes up to 36 percent. With an information value of 0.8, the probability gets to be 50 percent. At 0.8, the total number of identified customers is reduced from 430 down to 250, with only 23 at-risk customers going unidentified. A big improvement!

This is how we evaluate if a health score is working: Does it accurately focus resources where needed to effectively minimize churn and grow customer lifetime value?

With a high information value, it absolutely does. If your health score has a high information value, it is based on the leading indicators of churn — and you can reduce the amount of unnecessary intervention while increasing the coverage of at-risk customers.

The foundation for building your health scores

When you move beyond intuition or subjective input for health scoring and start relying on data, be sure to evaluate the information value of the factors you include, and don’t forget to back-test your final scoring mechanism. Just as importantly, start to think about different scoring mechanisms for different segments of customers. The best place to start is to think about taking control of your own data, and then create truly individualized health scores to match the expectations of your very individual customers.

[aditude-amp id="medium2" targeting='{"env":"staging","page_type":"article","post_id":1777312,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,enterprise,entrepreneur,marketing,","session":"A"}']

Here are five specific steps you can take to build a solid foundation for customer health scores:

• Step 1 — Break it down: Most health scores contain multiple factors, and to understand what action to take, you need the ability to drill down into different components of the score.

• Step 2 — Look for trends and changes: Customer behavior changes over time, and some health scores do not reflect historical trends.

• Step 3 — Zoom out for perspective: If you sell more than one service or product subscription, then lumping them together might hide important problems.

• Step 4 — Put health scores in the context of your revenue lifecycle: By understanding where customers are in their journey, you can map actions to their expected business outcomes.

• Step 5 — Automate the right customer interaction at the right time: You don’t want to wait for someone to review data and decide how to respond, as a health score doesn’t pay off unless you take action on it immediately.

Every company is different and will have a different approach, but at any company, high-level insight into customer health is good, and automated actions based on customer behaviors are even better. Use flexible analytics to monitor customer health and automated playbooks to keep customers on track. And, with in-depth insight and timely information, you can take in-person action as needed to deliver continuous customer value and keep customers for life.

---

Ryan Thomas is SVP of Customer Success at ServiceSource.