Demand Forecasting for a Medical Devices Company
Demand forecasting using near real-time data is crucial to any enterprise’s supply chain management system and plays an integral part in developing scalable supply chains with reduced complexity. An optimized supply chain can significantly alter the quality of offerings, enhance customer experience, and create business value.
So, when a leading medical devices manufacturer wanted to build a scalable supply chain, optimize its production planning, and reduce loss from back orders, Mu Sigma created a highly scalable solution that was data and insight driven.
A leading medical devices manufacturer wanted to optimize its production planning and reduce loss from back orders.
At the macro-level, the client wanted to:
• Improve the accuracy of its existing demand forecasting model to increase supply chain efficiency
• Create a demand forecasting framework for high demand-stream volatility segments with high-levels of accuracy
The client’s past efforts to deal with the problem involved working with analytics enablers that adopted a black-box approach to problem solving. With such contained approaches yielding no results, the client wanted more visibility and insight into the problem-solving methodologies.
The Mu Sigma Approach
To solve a problem of this scale, Mu Sigma leveraged its Art of Problem-Solving frameworks to understand the underlying problem DNA. We began by interacting with various stakeholders across the client’s supply chain and leveraged our internal framework, muUniverse , to map the complexity of the problem space and understand the interconnectedness between various problem areas such as demand planning, backorder management, production planning, inventory management etc.
The root of the problem stemmed from their difficulty in forecasting demand for their entire portfolio at highly accurate levels. This had a cascading effect leading to plant inefficiencies and sub-optimal inventory levels.
We followed a comprehensive methodology to enhance existing demand forecasting model accuracy to ensure short term accuracy and long-term volume alignment.
Mu Sigma created an advanced analytical framework aimed at improving the client’s demand forecasting. We followed a four-step approach equipped with a feedback loop to create a continuously improving model leading to high forecast accuracy.
• Developed an anomaly detection framework
• Analyzed the variations across multiple hierarchies of product, geography and time
• Ensured quick actions and regular monitoring
• Set up Time Series via application of 14 different statistical model
• Model Selection was done with eminent accuracy
• Developed an Ensemble model for short term volume alignment
• Forecasted demand for the next 2 years, enabling improved planning and revenue projections
• Ensured regular interaction with demand planners to incorporate business inputs
• Leveraged differential modelling strategy to add a BI layer
• Established systems to automate and scale the feedback process
• Tracked the Statistical forecasts and performed Root Cause Analysis to identify causes of forecast errors, if any
• Developed frameworks for benchmarking & performance measurement
• Helped optimize the process for identified focus areas
The statistical solution designed by Mu Sigma was a fine blend of scalability and accuracy. The solution was augmented with a feedback loop from demand planners which helped improve short-term as well as long-term forecast accuracy. With this, we enabled high-value outcomes for the client such as:
• Improved forecast accuracy from 53% to ~70% with nearly 95% accuracy across stable segments
• Improved overall process efficiency
• Reduced lead time and operational costs
Click here to know about how Mu Sigma helped a leading retailer improve labor forecast with an accuracy of 95%.