Dick Smith Electronics

Dick Smith's forecasting ability improves with SAS

Dick Smith Electronics, one of Australia's most recognised retail brands, has come a long way since 1968, when the founder Dick Smith opened a car radio installation business in Sydney. The company is now owned by Woolworths and offers a wide range of consumer electronics such as computers, mobile phones, fax machines and audio. More than 3500 people are employed within the 320 plus company-owned stores based around Australia and New Zealand.

The success of the company has led to it now offering more than 30,000 products and this has created one of its greatest challenges - that of accurately forecasting demand. With such a large number of products available, inaccurate forecasting means that shops selling popular products may find they run out of stock, while overstocking on slow-moving stock takes up too much shelf space and ties up inventory dollars.

Alastair Breingan is Dick Smith's Business Process Re-engineering Manager. He says a difficult issue with forecasting is the intermittent sales pattern of items. "There are some products, especially the very technical ones, where we might only sell one a week, whereas for others we might sell thousands a week," he says. "This means having accurate sales forecasts is very important. We had an old in-house forecasting solution but we were getting increasingly frustrated with its high error rate of more than 60 per cent."

A chance meeting with SAS last year to discuss the benefits of a number of its other products led to Dick Smith being given a development release of SASŪ High-Performance Forecasting solution. This solution allows users to automatically generate large volumes of forecasts to assist with their planning requirements. It is capable of generating millions of forecasts very quickly to reveal trends and predict future fluctuations in sales.

"We spent two months testing the SASŪ High-Performance Forecasting solution that SAS provided and were so impressed with the results we signed an agreement to purchase it in May this year," Breingan says.

To test the solution Breingan used the previous year's sales data to see how accurate SAS' forecasts would have been if it had been used at the time. They showed a forecast error rate of just under 25% which is a significant reduction. He was impressed with the results as well as the solution's ability to integrate into Dick Smith's existing systems and run as a stand-alone system.

"Many forecasting solutions require some form of human intervention but we didn't want that," Breingan says. "Many of our processes are heavily automated and the SAS solution allowed us to run our forecasting in the same way."

Dick Smith uses the solution to forecast sales of around 15,000 products that traditionally have intermittent sales patterns. Each week, information is pulled down from the company's AS/400 to desktop level and crunched. The data is then fed back into the AS/400 and integrated into the company's inventory system.

"The forecasts are generated in batch mode in a process that happens very quickly," Breingan says. "While there is a lot of pre-processing time required, the actual forecasting is very quick - it only takes a few minutes. We instruct the system to run a variety of forecasts and it provides us with the best model."

The solution is set to provide two forecast scenarios: one is a weekly forecasting out 12 weeks into the future, and the other is a monthly one forecasting out 6 months. Some of the key benefits that have resulted from the implementation of the solution include the ability to better match demand for products with inventory.

Dick Smith has just received the next release of the SASŪ High- Performance Forecasting solution. "The new release has many new functions including the ability to include data such as whether a product is being included in a promotion," Breingan says. "This will further increase the accuracy of the forecasts."

Breingan adds Dick Smith is looking at other SAS products particularly in the data mining and loss prevention area. "The success we are having with SASŪ High-Performance Forecasting solution means we are keen to try its solutions for fraud and shrinkage," he says. "We have been impressed not only with the solution but with the excellent support that SAS provides. We have also had direct access to the Research and Development facilities in Cary, where we have provided valuable input into future releases of the software. This is excellent service."