Sales & order forecasting for the food services company

To solve for

• Sales & orders forecasting is done monthly at aggregate level and at store level(1800+) manually. Monthly manual forecast is built which is time consuming & error prone.
• The accuracy of the manual forecast is around 35-50% at store level. Forecasting at daily, hourly and half an hour is required for variable manpower model but not possible building manually.
• Also, other forecasting is required such as channel, menu, category wise sales & order forecasting which is not possible building manually.

Solution

• Built Deep Learning model for forecasting the sales and orders at the accuracy of 95% and 97% respectively at aggregate level and more than 80% at the store level.
• Forecasting takes only few hours as compared to couple of weeks.
• An AI based conversational virtual assistant is provided that can answer any sales and order related query asked in natural language.

Benefits

• Improved Sales & Order forecast accuracy is helping in better marketing, supply chain and manpower planning with less wastage.
• Forecasting at 30min help in preparing accurate variable manpower plan.
• Inefficiency cost reduction in the supply chain & manpower planning.
• Self-help: Conversational virtual assistant provide answers to any sales query.