From raw transactional data to retention strategy. This project covers a full customer analytics pipeline built on real B2B data from a machinery rental company — exploratory analysis, RFM segmentation, churn prediction, and CRM recommendations.
Machinery rental companies live and die by repeat business. Losing a customer doesn't just mean one lost contract — it means losing a long-term revenue relationship that's expensive to replace. The challenge: with hundreds of B2B clients in the portfolio, it's nearly impossible to know which customers are quietly drifting away before it's too late to act.
Transactional purchase history from a B2B machinery rental company — covering 1,606 customers across multiple product categories, time periods, and seasonal demand cycles. The analysis included purchase frequency, recency, revenue contribution (Monetary), product mix, and temporal patterns.
The project had four components working together:
| Part | Focus | Key Output |
|---|---|---|
| 1. Exploratory Analysis | Understanding the business | Fleet, pricing, seasonality insights |
| 2. Customer Segmentation | Who the customers are | 5 RFM segments |
| 3. Churn Prediction | Who is at risk | ML model, AUC ~0.72 |
| 4. CRM Recommendations | What to do about it | Actionable retention playbook |
What a Client Can Do With This
The high-risk segment identified by the churn model represents customers who look active but show behavioural patterns consistent with pre-churn. If a sales or CRM team targets this group with a proactive reactivation campaign — a check-in call, a loyalty discount, or a contract renewal incentive — even a 20% save rate on that segment translates directly to retained recurring revenue that would otherwise silently walk out the door.
The Bottom Line
This is the initial analysis for a prioritisation engine in a CRM system of a company. Instead of treating every customer the same, a machinery rental business can finally answer: who needs attention right now, and what should we say to them?
Related Resources
To see the code related to this analysis please refer to:
https://github.com/baharmar/Marketing_Analysis_Projects