Demand Forecasting in

manufacturing for a Fortune 100

Sep 2019, Sydney

Eamonn Barrett

Increase stock availability through accurate demand planning in a dynamic

environment

Challenge

Threefold: Short product cycles, long supply chains, majority of sales concentrated in a small number of customers

Solution

An ensemble of probabilities, econometrics and Artificial Intelligence delivered an accurate forecasting method

A manufacturing client wished to improve their forecasting approach by utilising advanced artificial intelligence. Traditional methods had struggled to perform in environment required balance between the pressures of long supply chains, short product life cycles, and the majority of sales concentrated in a small number of large customers, make for a complex and dynamic forecasting environment. In such a complex and environment, previous approaches had an unacceptable degree of inaccuracy- with MAPEs sometimes reaching 150%.

 

Remi AI was engaged on a brief that sounded much simpler than it was: develop a better forecasting solution. The Remi data science team leveraged concepts from probability theory, econometrics, and artificial intelligence to develop a forecasting engine,The engine ingested more than just sales history and product attributes - the client had rich CRM data that provided an indication of upcoming product demand using the sales pipeline.

 

Results

The Remi team was able to improve accuracies by up to 80% - a fantastic result. The outputs of these forecasts were built into the client’s manufacturing orders, providing unprecedented insight and efficiency to the operation.

  • Accurate forecasts

  • Contextual data

  • Increasing efficiency