Collaborative Forecasting Running Smoothly at Brooks Sports
Brooks Sports designs and develops high-performance running footwear, apparel and accessories which are sold in 80 countries worldwide. In 2001, when the company shifted from a broad product line to focus on high-performance products targeted at serious runners, it was clear that the forecasting process needed to change to support the strategic direction of the company. The existing forecasting process, based entirely on the judgment of the sales team, was limiting the company's ability to grow.
The strategy shift created a number of forecasting challenges for Brooks including:
- Inconsistent style growth: the new line of products experience growth rates anywhere from 0 to 50 percent annually.
- Long production planning horizon coupled with short product life: production and capacity decisions are typically made 18 months before a style is launched, average lead time for a style is 6 months and the product life of Brooks' styles range from 6 to 24 months. This means that planners must sometimes set the entire demand plan for a style prior to ever receiving a customer order, underscoring the importance of accurate forecasts.
- Increasing "at-once" orders: "at once" orders, which are placed for immediate shipment, historically accounted for less than 20 percent of total sales. Since 2001, however, "at once" orders have increased to nearly 50 percent of total sales.
- Evolving size curves: with its new focus on serious runners, the standard footwear size curve would not adequately reflect distribution of sales by sizes.
- No exposure to retail sell-through: the high-performance products are sold primarily through independent specialty stores who don't have the capability to share sales data with vendors.
With a corporate mandate from senior management emphasizing the importance of creating accurate and timely forecasts, Brooks completely revamped its forecasting process. An independent forecasting group, reporting directly to the COO and CFO, was established to coordinate input from various groups-sales, marketing, product development and production-and to remove bias from the forecasting process.
The forecasting group established a collaborative forecasting process with three primary steps:
Step 1: Produce monthly statistical forecasts at the SKU level to capture level, trend, seasonality and the impact of events based on historical data. Brooks chose Forecast Pro to create these forecasts due to a number of features available in the software:
- Ability to create accurate forecasts
- Flexibility to choose forecast models or let software automatically select models
- Capability to model events (particularly important for predicting spikes in demand with new product launches)
- Support for multiple-level models to produce consistent forecasts at all levels of aggregation
- Powerful override facility to enable collaborative forecasting
"Forecast Pro has been a great solution for Brooks," says Tom Ross, Financial Analyst. "Implementing Forecast Pro's event modeling is very simple, which is an essential feature for us because of our moving product launches. We also use event models to address the challenge of forecasting events that don't occur on a regular basis-such as races-which can have a dramatic impact on the sales of specific products. Another powerful feature of Forecast Pro is the ability to forecast a product hierarchy. This helps us to serve our multiple constituents within Brooks-we review higher-level forecasts with management and easily generate detailed forecasts at the SKU level for demand planning."
Step 2: On a quarterly basis, get sales management and sales reps to forecast sales for a 12-month horizon, focusing on major accounts. This input is gathered via the Web and then aggregated by the forecasting group.
Step 3: Compare the statistical and judgmental forecasts, and make adjustments to create the final monthly forecast. Ninety percent of the final forecasts are the same as the statistical forecasts-changes are most commonly made to the forecasts for new styles where the sales organization has important knowledge to add. These final forecasts are then automatically fed into Brooks' ERP system.
"Forecast Pro allows us to easily apply judgmental overrides, which is critical for us," notes Ross. "We now can systematically track changes, giving us a better understanding of our forecasting performance."
The commitment to forecasting has paid off at Brooks. Forecast accuracy has improved on average by 40 percent, unfulfilled demand has been lowered from approximately 20 percent to less than 5 percent, and closeouts have been reduced by more than 60 percent. The improved forecasting has also helped to smooth out production, resulting in lowered costs and better margins.