Up until two years ago, DIRECTV used Microsoft Excel to create four-month forecasts for 120 parts that are forward-positioned at 250 stocking locations, says Sullivan, manager of supply chain planning and forecasting. “The forward stocking locations receive inventory once a week, so between the number of parts, FSLs and weeks, we had about 500,000 different forecasted values.”
While this was a very complex Excel document, it still only projected an aggregate number of parts for DIRECTV’s three distribution centers and only at a monthly level – nothing at the local level and nothing by week.
Since then, the company has completed an SAP implementation, which has enabled the forecasting process to be automated. “We now are able to forecast at the local level and by week. We also are able to use historical data to identify trends and seasonality,” says Sullivan. He notes that demand at DIRECTV follows a strong seasonality pattern, peaking when the NFL season starts.
“We now are able to incorporate that detail into our forecast along with whatever promotions the sales and marketing teams are planning. So our journey was going from this Excel version through requirements gathering and determining what we needed, to the ultimate development and testing of a system.”
Getting a system up and running was only the beginning of DIRECTV’s journey, however. “Once we had the system working, we needed to learn how to tweak and adjust it,” says Sullivan. “We had outsourced the requirements gathering and development of the system, but those companies left when the system was delivered. That’s when the real work began.”
Sullivan offers two takeaways for companies embarking on a similar journey. The first is to set expectations upfront. “We actually thought we had done a very good job with that,” he says. “We met with the procurement people who would need our forecast and with the group that distributes the equipment. But it turns out that we didn’t do a good job of setting expectations with them.”
He notes that the prior Excel process, despite its limitations, had a certain level of accuracy. “When we implemented a system that gave us far more detail, the downstream folks didn’t see the accuracy in that. They would just dive into their weekly, local number and say it didn’t look right.”
Additionally, there were upstream failures, he says. Finance and sales groups did not understand the importance of their inputs or would just forget to input data, he says – another failure to set expectations upfront.
The other key takeaway was around simply accepting the number from the SAP demand planning program. “It really took us quite a while to dive in and realize that our vendor had not set things up right,” he says. “We had good information going in, but the settings were not right so the numbers we were getting were off. This made me really understand that with automation you are not dealing with a crystal ball but a black box. It is just doing calculations and it has to be set up right.”
Now that all the kinks are worked out, DIRECTV is happy with the results. “We have system that is running well, forecasting at a level of granularity we never had before and with a good level of accuracy,” he says. “And all forecasting values now are pulled automatically into the system.”
Additionally, the company has reduced incidences of parts shortages “almost to none,” Sullivan says.
“We also are working, with some success, to reduce our total inventory carrying costs,” he says. “In a typical day we carry $350m worth of equipment in warehouses and reducing that will result in real direct-dollar savings.”
