They Know What You'll Buy Next Summer (They Hope)

During the boom-boom 1990s, the California casual dress style, popularized by legions of silicon-collar workers, conquered corporate America. Cadres of khaki pants strolled through office halls, and companies instituted casual Fridays. Executives took to the style, too, with Bill Gates, among others, appearing on TV looking very chino. And San Francisco-based clothing retailer Gap led the charge to outfit them all.

Sales rose (besting $13.6 billion in fiscal 2000, up from $11 billion in 1999) as Gap opened Old Navy, took its Gap and Banana Republic brands global and entered the Web channel. But it turned out that keeping up with demand patterns, planning the merchandise mix in all those stores (now more than 4,100 under three brand names) and determining when stores should receive shipments was a nightmarish task. Gap's sales continued to grow with its number of outlets into 2001, but net income dipped last fall as the economy entered tougher times, and red ink dripped. Gap posted an $8 million net loss for 2001, and said it would close distribution centers in Kentucky and Holland.

2001 was Gap's toughest year. But long before reports of a big retail sector slump, Gap executives had committed to upgrading their ability to manage inventory worldwide. "We have a tremendous need to be more accurate in our planning, forecasting, allocation, pricing and optimization," says Ken Harris, Gap's CIO.

The retail industry has always been ripe for good forecasting technology that accounts for the thousands of products mass merchants carry, the myriad stores they operate, the impact of promotions on sales and the problems merchants can encounter with suppliers.

Retailers have collected vast amounts of point-of-sale data in data warehouses for years, but they haven't had the means to apply it effectively to their planning and buying because up until a few years ago, no computer or software application could process all of that data. The forecasting applications available today offer retailers better results because they incorporate more than just historical sales into their forecasts. Technology that in the past would generalize inventory needs across many stores now allows big-chain retailers such as Federated Department Stores to make sophisticated decisions about the needs of each of their outlets. "These new methodologies scale to handle much higher volumes of store SKU [stock keeping unit]-level forecasts, and they're also able to take into consideration a broader range of causal factors, such as price, consumer response and the promotional context of a sale," says Greg Girard, vice president of retail application strategies for Boston-based AMR Research.

That's not to say there's a fresh, clear path to profits in that traditionally brutal market. In fact, Bob Muller, vice president of inventory management at KB Toys in Pittsfield, Mass., says a new planning system that better spells out the needs of each of his company's 1,400 stores has definitely improved the allocation of merchandise. But external factors, such as consumer confidence declining after the Sept. 11 terrorist attacks, can make tabulating the benefits of forecasting a tough task. "With the recent events [of Sept. 11], it's obviously much more difficult to measure the impact on sales from the [calculations about] our inventory that we've done," he says.

Still, in a downturn, forecasting technologies are vital tools for retailers looking to stock their shelves to meet reduced demand. They help stores and warehouses carry less inventory and thus incur fewer costs. As the strategic investments by Federated, Gap, Sears and others demonstrate, retailers are digging in to fight for long-term gains.

Filling in Gaps

at gap, corporate buyers did not have a forecasting engine to crunch sales data and help predict how many boot-cut jeans to buy or how many would sell. Instead, Gap's planners and buyers had to gather current and past sales information from different, siloed planning and allocation systems and analyze it on their owna slow and tedious activityto determine what to buy, how much to buy and when clothes should arrive at stores. Often data was duplicated among those different systems, which affected the accuracy of their forecasts and decisions.

Harris, Gap's CIO since September 1999, says he recognized that the planning and optimization tools used by airlines could help his company. His search for a system robust enough to handle all of the sales data from Gap's 4,000 outlets took until March 2001, when Gap began implementing planning and forecasting applications from Minneapolis-based Retek, a retail software vendor. Harris expects to be fully up and running on a merchandise planning and forecasting system as well by the end of 2002.

"We needed this ability to plan and forecast in the same tool, have visibility into our end-of-season and preseason planning, our inventory position, and be able to match that up very quickly to our receipt planning and make adjustments [as necessary]," says Diane Silver, Gap's vice president of IT.

With the system from Retek, Gap will have a common set of tools for all activities across all divisions. It will integrate preseason planning activities such as determining how many faux-fur denim trench coats to stock in New York City stores with end-of-season activities such as clearance pricing. The tools also will help planners gauge shoppers' reaction to repricing items and speeding up or slowing down shipments of clothes, according to Michael Barrie, Gap's vice president of planning and forecasting systems.

And the new system will give merchants and planners visibility into each other's plans, into item-level and class-level forecasts, and into the financial pulse of the business, says Barrie. He says that having that visibility will help merchants and planners make better decisions about what to allocate, when to mark down and how much to mark down.

"Every item in a store is an investment," says Barrie. "Given the size of our company, those are very sizeable investments, and we want to get the best possible return on each of those investments," he says. So if stores receive shipments too soon, then clothes remain on shelves longer and the company's inventory costs climb while its return on inventory investment decreases. "The better you can make [shipments] just-in-time, the higher your gross margin return on your inventory investment [will be]," says Barrie. Silver says the new system will forecast at a very detailed style-color level based on a whole series of factors such as how fast an item sells, the date it gets put on a shelf in a store and seasonal sales patterns.

Having an integrated system with "a single version of the truth," she says, will let merchants and planners focus on decision making rather than gathering information from disparate systems. Also, merchants and planners entering the system through software on their desktops will be able to see the current status of each others' plans and forecasts. That visibility, says Barrie, will lead to better collaboration and decision making.

"It won't help us pick better product," says Barrie of the system. After all, buyers still have to have a keen sense of what's hip. "But it will facilitate more efficient decision making," he adds.

Better Than Your Average Meteorologist

Tough times hit Sears, Roebuck & Co. last October. The 115-year-old company laid off 4,900 workers and announced it would change the layouts of its stores to better compete with discount retailers. The company's 2001 revenues were down 3.6 percent from the previous year. Sears's CEO Alan Lacy promised to double his company's profitability during the next three years and reduce costs by $600 million by ameliorating order management, using retail and warehousing space more efficiently, and by improving inventory controls. A weather forecasting application will play a critical role in executing Lacy's strategy.

Since 1995, Sears, based in Hoffman Estates, Ill., has received long-range weather forecasts from Planalytics, a Wayne, Pa., software vendor. Those reports complement other traditional planning methods such as economic forecasts, company decisions about new or discontinued products, store openings and closings, and competitors' moves.

The weather becomes an important data point, says Jonathan Rand, director of merchandise planning and reporting at Sears. "If the weather impacts traffic in the store, it's going to impact your business," he says. Sears has added to its weather forecasts from Planalytics, a Web-enabled decision support tool from the same company that helps the retailer determine what actions to take to prevent overstocks and understocks based on the forecasts.

For instance, when Sears did its inventory planning for December 2001, it took into consideration that Planalytics had predicted that month to be warmer than normal in the Northeast. Indeed, during early December in Boston, temperatures registered in the 70s. Planalytics also suggested that the milder weather would have a negative impact on the sale of winter coats. To mitigate that risk, Rand says, Sears decided not to buy as many heavy coats in order to avoid taking markdowns at the end of the season. He says the company also decided to change its assortment from heavy-weight wool coats to lighter-weight leather coats.

Rand says that Sears cut back on its inventory buys because of the weather forecast. Although topline revenue suffered as a result of those scale-backs, Sears's carrying costs were lower because it was holding less inventory. "Our overall profitability was better because we didn't have a lot to mark down," he says.

Merchandise planners also realized that the mild December weather would negatively impact the sale of snowblowers, an item the retailer sold out of during the winter of 2000. To boost snowblower sales, Sears advertised zero percent interest rates for six months on snowblowers on the front page of its Sunday circular during September. Sears bet that consumers would remember the previous winter and that they wouldn't want to be left in the cold sans snowblower. They were right. Snowblowers were the top-selling hardware item that month.

In the past, buyers used the average number of air conditioners Sears sold during a particular month in previous years to determine how many air conditioners to buy for the current year. Then they'd look at the federal government's long-range weather forecast or the Farmer's Almanac to get a sense of how the weather would influence consumer behavior. Rand says those forecasts weren't very accurate. Finally, the buyers would bet on the number of air conditioners they thought they'd sell. To make sure they didn't run out of merchandise, Rand says, buyers would bring in products earlier in the season, even if the season was going to start later than normal. "If you know the season is going to start late based on the forecast, you won't bring in as much inventory as you normally would and therefore you won't have the carrying costs associated with that," he says.

Though the technology provides valuable information and insight, there's still the human factor: Buyers at Sears still have to make the final decision about what they're going to buy and how much. So even though the technology is accurate in its forecasts three days out of every four, there's still margin for error. "Have we used it as well as we should? Absolutely not," says Rand.

Rand says the difficulty of using the tool lies in the sheer complexity of trying to predict sales. He also says that local store managers accustomed to using historical sales data to make their buying decisions are reluctant to start using a software toolespecially weather forecastswhen they are already familiar with the difficulty their local meteorologist has predicting weather for the next five days. To overcome that resistance, Rand says, he tracked its accuracy (74 percent correct during 2000) and shared those results with the merchants. "When you see how accurate it has been, people say, 'This is darn good,' and then they start using it. That's what's happening," he says. Rand declined to share ROI figures but said the technology has more than paid for itself.

One Size Does Not Fit All

It used to be that Cincinnati-based federated Department Stores figured it had the formula for an ideal store. The model said that every Macy's should stock the same number, say six, of a certain style of men's shirts and keep those levels consistent throughout the year. The systems in place regularly reordered 10 percent of staple items such as socks, underwear, hosiery, cosmetics, men's dress shirts and cookware.

It didn't work. Sometimes, consumers couldn't find items they wanted and Federated lost potential sales. Other times, goods sat vigil waiting for shoppers who weren't interested, and Federated wasted money on unproductive product displays.

"Six is not the right number [of shirts] 52 weeks a year. Our business has peaks and valleys," says Gale Weisenfeld, vice president of retail technology at Federated Merchandising Group. Federated needed a system that reacted to those seasonal changes in demand and that could handle a larger number of items for its more than 450 stores, which include Bloomingdale's, Burdines, Goldsmith's, Lazarus, Macy's, Rich's and The Bon Marché.

1 2 Page 1
Page 1 of 2
7 secrets of successful remote IT teams