scroll

A leading manufacturer of medical devices and surgical products in India, including sutures, surgical mesh, clips, and appliers, catering to over 2,500+ hospitals and nursing homes across India wants to plan their inventory to reduce their working capital.

We made a visible and measurable impact to our client's business

30%

Reduction in inventory leading to reduction in working capital while maintaining the current fulfilment rate.

Challenge

Industry Overview

The earlier forecasting method was not scientific, leading to inconsistent forecasts and poor accuracies. A static inventory norm for all SKUs (say 15% of last 3-month average sales) resulted in high working capital due to high inventory and low fulfillment rates. Inventory planning for 8,000+ SKU-stock point combinations add another dimension of complexity to the problem.

Problem Overview

In case of this company, the highly customized nature of the SKUs also meant approximately 1000 SKUs in their portfolio being supplied from their nine stock-points. While about 150 SKUs contribute to around 77% of their sales, the long tail SKUs are essential for maintaining products’ assortment.

The company aspires to.

1.Establish a robust and scientific mechanism for estimating Demand for the SKUs at each of the stock-points.

2.Avoid overstock, stock-out situations, and to counter variations in Demand by optimizing inventory tailored for the SKU at a stock-point.

3.Understand the shifts in customer-mix.

Why were we brought in?

The company asked Ganit for help in building a solution using advanced ML algorithms that would help Lotus bring about a reduction in the working capital.

Our approach

Methodology

We used the following data sets for estimating the demand.

1.Demand data.

2.Item metadata.

3.Extraneous Features.

The team classified the SKU-Stock points into an ABC-XYZ framework – A combination of the sales contribution and ease of prediction.

Ganit's team performed multiple treatments to the data, eventually yielding the raw data containing the SKU-Stock point combinations with the actual demand.

Forecasting

We estimated Demand at SKU – Stock Point level for 1,563 combinations, contributing to 83% of the total Demand. Multiple forecasting techniques were attempted, and we opted to use AWS Forecast for the following reasons.

1.Improved accuracy and better response to demand variations.

2.Flexibility to adjust the forecasts for controlling forecast bias.

3.Flexibility to input complex relationships found in the data in the form of regressors.

4.Running multiple threads in parallel.

5.Quick setup.

Data Pipeline and Workflow orchestration

Cron job was scheduled to automate forecasting activity.

A valuable difference

Impact

Using Ganit's Solutions, the company gained the ability to convert their data into strength - the mechanism, while providing demand forecasts and supporting the supply chain, also ensures that the service levels (fulfillment rate) do not dip below the 95% threshold for any SKU-Stock point.

Top