ML-Based Churn Forecasting for Energy Customer Base Planning

Problem and Context

An energy provider needed a more reliable view of its future customer base after the energy crisis increased planning uncertainty and sharpened the need for better demand control. The client wanted to improve controlling assumptions for energy volume planning and set sales acquisition targets on a more objective basis. A central challenge was forecasting customer churn across different product segments, each with different behavioral patterns over the contract life cycle. External disruptions such as COVID-19 and the energy crisis added complexity because they materially influenced historical data but also needed to be reflected in a realistic future planning view.

<9.4 pp

average error in customer churn prediction

3

core churn behaviors identified

>85%

of churn cases linked to distinct customer behaviors

Approach and Solution

The project combined transparent baseline logic, business-oriented forecasting, and scenario simulation into a planning capability for future customer development. Churn patterns were first reviewed across the customer life cycle to create reliable benchmark views before the forecasting approach was refined further. Rather than excluding unusual periods, the team explicitly reflected major disruptions such as COVID-19 and the energy crisis in the planning logic. The final output was embedded into a configurable simulation tool that allowed business users to test scenarios for customer inflow, market price levels, time horizons, and product-level aggregations.

Results and Impact

The initiative gave the client a stronger basis for planning future customer volumes and related energy demand while making churn assumptions more transparent and scenario-driven. By combining benchmark logic, improved forecasting, and simulation, the solution strengthened the objectivity of controlling and sales target setting. It also enabled more flexible what-if analysis across products and market conditions, helping teams assess uncertainty and sensitivity in customer base development with greater confidence.

Want to learn more about this case?

Contact:

Role:

Email:

Steffen Illig

Partner, Project Manager and Expert for Data Analytics

steffen.illig@5v-strategy.com