Waitrose puts new forecast system in its basket

Waitrose is rolling out a forecasting system to help it better predict demand for goods from customers.

The retailer - which has 184 stores that carry up to 25,000 different product lines each - decided it had to improve its forecasting system and carried out pilot projects last year.

Waitrose product systems manager, Mathew Hallett, explained predicting what shoppers want is a tricky business because retailers have to "know what the customer wants to buy, days before they buy - even though when they walk into the shop they don't know what it is".

silicon.com Retail & Leisure

Get the latest retail and leisure news straight to your inbox. Sign up for the R&L newsletter today!

Gail Richmond, the retailer's manager of branch ordering development, said while the previous forecasting system could react to events, it "didn't see things coming".

This meant wastage, loss of sales and unhappy customers, she said. "We knew there was a requirement to deliver a demand forecasting system for all our stores."

Waitrose is using technology from SAS for its forecasting engine, to help calculate demand and monitor performance, while other parts of the new system have been built by the retailer using Java.

Last year the supermarket ran several trials of the technology at stores and is now in the process of rolling it out to all its branches.

The system rollout has seen a reduction in wastage by between three and 20 per cent in some areas, halved the amount of orders amended by the store and reduced the amount of stock in stores. Speaking at the SAS Forum in Stockholm, Richmond said: "It's very pleasing."

Richmond said there are now 41 branches live with the system. The rollout across all stores is due to be finished by October. Waitrose is also exploring similar improvements to its warehouse forecasting system and looking at sharing information with suppliers.