Why I Overestimated Predictive Health Scores – and What I Learned

A few years ago, I felt like I had discovered the true value of predictive health scores. I believed these scores would be the ultimate cure for churn prevention and that a clever, predictive score was exactly what companies needed to anticipate challenges and prevent customer churn. My team and I invested significant time and resources into developing such a model.

Looking back, I think it was a mistake.

Not because predictive health scores are inherently useless – they can certainly make sense in specific scenarios. What I underestimated was the complexity of individual customer requirements. It quickly became clear that very few companies were capable of developing a truly accurate and predictive score that works across diverse customer situations. Our model failed repeatedly, especially with complex use cases or in complex markets.

But you learn from mistakes – and this exact realization fundamentally shifted my mindset. Instead of obsessing over an ideal "predictive" model that is often too static and isolated, I now focus on something far more important: actual value for the customer.

Value Trumps Predictive

What I underestimated was the importance of the value journey. Companies that want to create real value for their customers must dive deep into their use cases, understand their challenges and tailor solutions accordingly. A "health score" attempting to predict the state of a customer relationship using historical, purely usage-focused or isolated data frequently neglects the changing conditions and needs of the customer.

Today I rely on value-oriented strategies that align much more closely with real customer needs and goals. The idea is to clearly define target value together with the customer and use this value as a guide for every subsequent interaction and decision. This means the health score is not a central goal, but merely a tool on the path toward a bigger picture.

The Key Question: What Drives Value?

Ultimately, one question matters above all: How can we help our customers achieve their own goals? This requires a high degree of collaboration and tight alignment with what the customer truly wants and needs. It is about understanding where the customer stands, where they want to go and how we can support them along the way.

Predictive models might be useful in some cases, but they are only a small piece of the puzzle. What makes the difference is a deep understanding of customer use cases and the ability to shape the path to success together with the customer.

My takeaway: Spend less time investing in complex predictive models that often fail anyway, and focus more on the customer's actual value instead. In the end, that pays off – both for the customer and for the company.

Want to learn how to build your own Value Framework? → Book a free Value Assessment.