Leveraging enterprise data and advanced analytics in core operational processes:
Demand forecasting at Cisco
This case was prepared by Blake Johnson of the Management Science & Engineering Department at Stanford University. The author gratefully acknowledges the contributions of Anne Robinson and Andrew Fisher.
Introduction Accurate forecasts at the product level are critical to many core business processes across a company, from financial planning to sales and marketing, supply chain management, and customer service and support. However, generating good forecasts is challenging for a number of reasons. First, relevant data includes both large enterprise data sets about the present (current customer orders) and the past (historical demand), as well as forward-looking insight from sales and marketing, both which can be difficult to access and effectively utilize. Second, most companies have a diverse range of business units, products, and customers, ruling out "one size fits all" forecasting approaches. For example, Cisco has over 18,000 product IDs, ranging from high-end enterprise systems to consumer products, which it sells to a diverse set of customers around the world.
"The top two challenges are speeding up our new product introduction to world- class levels and forecasting accuracy & demand management." Angel Mendez, Senior Vice President of Customer Value Chain Management
At the operational level of forecast process implementation and execution, the challenges are even greater. Different functional groups use different data definitions and different IT systems, and have different goals and objectives. For example, finance and sales and marketing focus on revenue and track the time orders are booked, while supply chain focuses on units and tracks the time orders are to be delivered. To create the best possible forecast all available data and organizational expertise must be utilized, and relevant stakeholders engaged in the process of forecast generation, review, and sign-off. To realize the value of the forecasts generated, an even wider range of stakeholders across the company must be convinced of their quality, and must be able to easily access them. Cisco solution Cisco has developed a forecast generation process that combines advanced analytics applied to enterprise data on bookings and historical demand, with a consensus process that incorporates current insights from sales and marketing and finance. To ensure broad organizational adoption, forecast performance is careful monitored and documented, and the forecast generation process is kept transparent and accessible. Finally, performance metrics and incentives drive on-going improvement in both forecast quality and utilization. Cisco's Demand Forecasting Process The development of Cisco's demand forecasting process was sponsored by Karl Braitberg, Vice President of Demand Management and Planning. The process was
designed to address three key challenges: 1) data acquisition...