A maze of markets
Our client, a global B2B supplier of chemicals, plastics, agro-sciences and advanced materials, was facing a serious challenge: customer churn. Despite their market-leading position across multiple industries and countries, sales had been steadily declining since 2019. While some churn was expected, the real concern was unexpected customer departures with no clear reason.
Internally, sales teams struggled with fragmented data, and there was no reliable way to predict or prevent churn. Our job was to build a churn model to bring clarity and equip sales teams with actionable insights. Given the complexity of the business, we took a phased, iterative approach.
We focused on two key business units: Direct Customer Solutions (DCS) and Plastic and Specialty Packaging (PNSP). These were chosen due to strong stakeholder support and their strategic revenue importance.
Our goals:
- Develop a predictive churn model for these units
- Create a scalable framework for future expansion
Flying the flag
First, we tackled the data challenge. With sales data scattered across multiple systems and accounts, spotting churn patterns was tricky. Instead of jumping straight into modelling, we built a churn flag dashboard – assigning risk scores to customer accounts based on key indicators. Integrated into monthly meetings, this dashboard enabled real-time discussions and action on potential churn risks.
Once this was in place, we launched the first iteration of the churn model, using historical sales, customer engagement, and supply chain data.
Although still early days, the impact has been significant. Sales teams now have a clear way to identify at-risk customers before they leave, rather than reacting after the fact. Feedback has been overwhelmingly positive, with teams feeling more confident in tackling churn.
A phased approach
Looking ahead, we plan to:
- Roll out the model across all business units
- Integrate it into the client’s CRM for real-time alerts
- Refine accuracy using feedback from sales teams
- Expand data inputs to enhance predictive power
This is just the start of a long-term, data-driven approach to customer retention, turning churn prevention from guesswork into strategy.