The high fashion industry faces major challenge in predicting sales of specific products lines. The problem is compounded by choice of colors and sizes. The costs of poor estimates of demand take the form of end-of-season markdowns for over supply and stockouts when demand has been underestimated. For example, one woman's parka sale exceeded the forecast by over 200% and another ended up at less than 15% of the original forecast. In the 1970's markdowns accounted for less than ten percent of the dollar sales in department stores. By 1990 that number had skyrocketed to more than twenty percent.

Sport Obermeyer is a leading manufacturer of fashion skiwear for both adults and children. The peak retailing season is two months and demand is uncertain. As its business increased in the late 1980s, Obermeyer found that it was running into serious manufacturing constraints during the critical summer months. At the same time competition was forcing all companies to increase their variety in order to stay competitive. These are some of the actions Sports Obermeyer took in response to the growing complexity of its manufacturing planning process.

• The twenty-five largest retail customers were invited to sneak preview in February the year before. Their early purchase orders were used to improve initial forecasts.
• The members of the buying committee made independent forecasts for different products rather than trying to reach a consensus. Products for which individual committee members differed the most were modeled as having the most uncertainty.
• Production over the manufacturing horizon was categorized as nonreactive (planned in advance) or reactive.
• Early production capacity was nonreactive and utilized with optimal production plans for products with less uncertain demand. In addition, some nonreactive capacity was used to produce minimum amounts for the more uncertain product lines. The mid-summer, critical production capacity was reserved for more uncertain demand items.
• Key design decisions were made that reduced variability and cut the potential for component shortages holding up production. For example, they increased the use of black zippers as a color contrast. This reduced the number of zippers by fivefold.
• Colors and materials were standardized so that a typical season there were two or three shades for each design cycle instead of five or six.

All of the above actions were coupled with sophisticated forecasting models, probabilistic inventory models and optimal production plans. As a result, the need for markdowns was dramatically reduced in an industry in which profits average just 3% of sales. Costs declined by 2% with a corresponding two-thirds increase in profits.