As any subscription-based business knows, customer churn can be a major brake on growth. Not only are customers expensive to replace — a common rule of thumb is that acquiring a new customer costs five times as much as keeping an existing one — but new customers can also be harder to cross-sell/up-sell to.

That’s why growth-minded businesses place extra emphasis on customer retention and existing account monetization. While many organizations have a good grasp on how they perform on retention or cross-sell/up-sell, a key challenge is knowing when a customer is likely to take either kind of action so you can respond accordingly.

Real-time, predictive analytics can be the answer to minimizing attrition, maximizing cross-sell/up-sell, and ultimately driving significantly improved revenue metrics.

How Churn and Cross-sell/Up-sell Affect Net Revenue Retention  

Net revenue retention (NRR) is one of the most critical metrics for any subscription business. NRR is calculated by adding your starting monthly recurring revenue (MRR) to the change in MRR (expansion less contraction and churn), and then dividing by the starting amount. The formula is expressed as: 

starting MRR + expansion – contraction - churn 
starting MRR 

Companies can significantly boost their long-term success and corporate valuation by improving their NRR.  

How significantly? According to our analysis, NRR explains almost half of a company’s variance in enterprise value to revenue multiples (see Figure 1). Each percentage point increase in NRR is associated with a 0.5x change in multiple.  

Gain Foresight on Customer Behavior

Most organizations have some way to track customer churn numbers at an aggregate level, and some may run analytics to explain why the churn happened. But this information is hindsight – it only gets you so far to truly moving the needle on NRR. It takes foresight to understand when churn may occur so the business can mitigate the problem before it happens. A similar situation exists with up-selling and cross-selling, where opportunities often go overlooked until they have passed. 

L.E.K.’s predictive, machine learning-driven models provide foresight on customer behavior so you can see how specific accounts are likely to perform (e.g., is Customer X at risk? Is Customer Y ready for an upgrade?) (see Figure 2). 

Build Intervention Events to Change Outcomes

Furthermore, knowing what customers are likely to do is only half the story. Knowing which intervention events will actually change an outcome are equally critical. By analyzing historical data and working with your Customer Success teams, we define and embed mitigation steps into your employee’s workflow. These propose tailored interventions for each account, and becoming increasingly smart over time as more data is generated and analyzed (Figure 3). 

Overcome Data Silos for Greater Insight

A subscription business has a treasure trove of customer data. But internal silos can prevent the organization from using all the relevant data at its disposal. Our model seamlessly combines data from multiple platforms as well as high-value external datasets to create a single view for analysis (see Figure 4). 

Harmonizing datasets maximizes the model’s predictive power and unlocks cross-functional insights (think product usage and how it affects optimal cross-sell/up-sell strategy). It also improves the model over time by continuously adding new data around problems to mitigate and opportunities to capture.

Improve Your Customer Retention and Profitability 

As important as it is to acquire new customers, it’s ultimately more effective and efficient to monetize the customers you already have. And that makes customer retention an equally important priority. With L.E.K.’s predictive modeling capabilities, you can gain both insight and foresight on how to optimize your company’s NRR for long-term success. 

For more information, please contact

Related Insights