KMWorld Best Practices, Oct 2002: O'Reilly Auto Parts meets demand for all seasons

By Judith Lamont

When O’Reilly Automotive went public nearly 10 years ago, it planned to grow, and over the past 10 years has seen nearly a fourfold increase in the number of its auto parts stores. Early on, the company knew its in-house inventory management system could not scale to accommodate that growth. In addition, O’Reilly sought a more sophisticated model for demand management. After considering the alternatives, the company selected the Demand Network Suite from Nonstop Solutions, which produces demand planning and inventory management software.

Demand management takes over at the point at which finished goods are ready to be purchased. It manages the flow through warehouses and distribution centers out through retail stores. Keeping inventory at optimal levels is at the heart of demand management. Enough product should be available to maintain customer service targets, without tying up too much money and space in excess inventory. Retail demand is measured by point of sale (POS) data.

Although demand management has some elements in common with supply chain management (SCM), it also has some important differences. Like distributors and retailers, manufacturers want to optimize their inventory, and can benefit from analytics that provide information about delivery response time and pricing. However, while manufacturers might have dozens of suppliers, a retail operation such as O’Reilly handles thousands of different finished products. Thus, projecting retail customer demand can be more challenging than managing the flow of raw materials.

“We knew we needed better mathematical formulas,” says Mike Williams, VP of Information Systems at O’Reilly, “especially calculations that take seasonal variations into account.” Although the existing system did recommended stocking levels, every item had to be checked manually. “We also knew it would be difficult to expand our product line using the present system,” adds Williams. O’Reilly, which serves both professional auto repair shops and do-it-yourselfers in markets in the South and Midwest, planned to significantly increase the number of products offered

At the time, the selection of Nonstop Solutions might have seemed like a risky choice. A relatively new company, Nonstop had recently changed its direction to focus on demand management for individual firms rather than on coordinating sales between groups of manufacturers and retailers. However, O’Reilly management was convinced that the mathematical model developed by its founder, Stanford University professor Dr. Hau Lee, would serve the company well. Moreover, Nonstop was interested and responsive.

The implementation process for the new demand management system went smoothly. The team from O’Reilly included both business and technical staff, each providing a different perspective and ensuring user buy-in. Nonstop Solutions was an active partner, taking time to understand the business and providing the initial rules for managing the inventory. “We started with small steps,” Williams notes, “gradually adding in product lines from each vendor.” Training sessions were started to show operational staff at headquarters how to use the system.

Now, the purchasing department at O’Reilly can review a list of recommended buys that is generated overnight. A daily agenda is produced that flags exceptions, such as unexpected spikes in demand, for closer examination. The list is prioritized so that the most urgent decisions are attended to first. As buyers in O’Reilly’s purchasing department became more comfortable with the system, they could opt to activate an auto-accept logic that generated purchase orders. The orders can be in electronic, fax or mail form, depending on the vendor’s requirements. As a result of those efficiencies, the number of SKUs handled by each buyer has increased from 40,000 to 100,000.

With more precise predictions of demand, O’Reilly was also able to reduce its inventory, which freed up $66 million in capital over the past few years. The company used some of the savings to purchase Mid-State Automotive Distributors in August 2001. That acquisition added seven states to O’Reilly’s roster, creating a territory of 16 contiguous states.

The demand management system can take in data from outside sources in order to further refine its predictions. For example, O’Reilly plans to integrate R.L. Polk’s (rlpolk.com) registration data, which includes location, model and age. By knowing the exact vehicles in the area, the company will be better able to anticipate the parts that might be needed. O’Reilly also plans to add data about failure rates of parts for different vehicles, because that variable affects demand. To allow for more elaborate analyses, the company is considering adding a business intelligence tool.

“The system has huge potential,” says Williams. “Once we began capturing POS data, we knew what was happening at the counter within 15 minutes of the sale. That sets the stage for us to do very current analyses.”

Another benefit of the modeling capability is better planning for new stores. O’Reilly can look at the demographics of an area such as an agricultural community and match them to those for existing stores in similar areas. When the new store opens, its stock already is a close approximation to the needs of that community. O’Reilly can then fine-tune the store’s product mix to meet specific local demand.

“The Nonstop Demand Network Suite is very adaptive,” points out Calvin Lee, chief scientist of Nonstop. “Business rules in the system can easily be changed if the historical data has been affected by unusual factors such as a harsh winter.” The model begins with a set of default values, and then customer-specific data such as handling costs or the cost of borrowed money are input as part of the implementation.

“The software makes the process look simple,” says Lee, “but the math going on in the background is complicated.” A process called demand cleansing adjusts the historical demands for events such as promotions that might skew the results. Customers can run their own simulations to recalibrate the system’s forecasting model, or re-run seasonal profiles as conditions change. Lee emphasizes that there is no single right answer for all companies with respect to inventory. “In the pharmaceutical industry, for example,” Lee notes, “a distributor might get a better deal in the long run by carrying a larger inventory because of forward buying opportunities due to vendor deals or increasing prices.”

O’Reilly began with a compelling and well-defined business need. The company selected software that improved and automated key processes, with built-in options for flexibility and growth. Unlike many IT implementers, O’Reilly has been able to identify a clear link between the Nonstop system and performance enhancement, with greater employee productivity and cost savings that supported additional growth. In Nonstop, O’Reilly found a committed vendor that was responsive to its requirements. All those ingredients add up to a solution that is likely to thrive, rather than to languish, as time goes on.

Judith Lamont is a research analyst with Zentek Corp., e-mail jlamont@sprintmail.com