Smarter dispatch of technicians
The outcomes
- New predictive model achieves 90% accuracy (versus 65% for the legacy approach).
- No unnecessary high-cost assignments.
The context
A leading telecom provider needed a smarter way to predict which technician to send to clients. The challenge: balancing cost and expertise. Sending an underqualified technician often led to rescheduled visits and unhappy customers, while sending an overqualified technician increased costs unnecessarily.
The logic
The client wanted a solution that could accurately determine the right technician for each job—optimizing customer satisfaction and operational efficiency.
The solution
We developed a predictive model using advanced analytics and data such as client location, type of residence (house, apartment, floor), and more. The model recommends the appropriate technician type:
- Profile 1 for simple tasks (least expensive).
- Profile 2 for moderately complex jobs.
- Profile 3 for advanced work requiring top expertise.
Key steps included data collection & cleaning, ensuring high-quality inputs for accurate predictions, and the building of a state-of-the-art decision tree algorithm to develop and validate a first model.
Next steps in applied excellence
The pilot is underway, and early results are promising. The next phase will focus on A/B testing and continuous improvement to further enhance accuracy and efficiency.