Case study card:
Demand forecasting and price optimisation
Primary Industries and Infrastructure
Inpost was looking for a company which was experienced in Demand prediction topics.
The main need was to create an algorithm that, based on historical data, would be able to forecast the daily sales volume for a particular customer (parcel lockers and courier).
InPost was not able to compete with its competition because of problems with meeting the demand and appropriate pricing strategy for individual sellers. That caused delays in delivery and customer churn.
The team built and implemented Machine Learning prediction models trained on historical data, macroeconomics factors and other 3rd party data. Rare events such a Covid were taken into the account.
The model predicts demand and calculates optimal price to a private sellers. Solution automatically forecasting monthly and weekly sales by individual products and companies. Based on the demand and supply analysis optimal price is chosen.
Solution helps to plan couriers (workforce) according to the future demand and takes into account their movement from point A to point B. It uses workforce coverage optimization (e.g. use of maps and geospatial data to allocate staff and salesforce to maximize coverage and efficiency).
The solution helps managers in better and more accurate planning, save time of manual work, eliminate constantly making predictions in excel and taking into account Covid-19 and macroeconomics factors which had a big impact on sales.