August 7, 2001

[The Wall Street Journal Interactive Edition]


Page One Feature

Retailers Attempt to Get a Leg Up
On Markdowns With New Software

By AMY MERRICK
Staff Reporter of THE WALL STREET JOURNAL

It's one of the great nail-biter decisions of retailing: just  when to hit the markdown panic button.

Discount chain ShopKo Stores Inc. got nervous this spring, as shopper after shopper shunned its stretchy nylon track pants.  They had a comfortable elastic waist. They had bright, contrasting stripes down the sides. But they simply weren't budging at their original price of $16.99.

ShopKo didn't want to resort to its strategy of years past:  whittle the price a bit here and there, then chop it methodically  once a month -- to $9.99, then $7.99, then $3.99. Instead, it took  only two measured swipes, marking them down to $10.79 on average in early May, then to $9.75 a month later. And it held firm there. With the new pricing plan, ShopKo sold out the last 12,500 pairs during the three-month clearance period, for a gross profit margin of 31% -- nearly double its forecast.

Airplane Seats and Bikinis

It wasn't a random decision. ShopKo is one of a handful of retailers, including J.C. Penney Co., L.L. Bean Inc., Liz Claiborne Inc. and Gymboree Corp., trying to perfect the  science of the markdown. They have been experimenting with  sophisticated new software programs to test principles similar to  "yield management," which airlines mastered years ago to eke out the maximum profit from every seat. Like a seat on a particular flight, an item such as a bikini is in demand for a limited time; as the end of the season approaches, its value to customers plummets. A big challenge: trying to outfox customers who have been more willing to wait and wait for a bargain.

Using number-crunching consultants, armed with mathematical  models pioneered by think-tank researchers and astrophysicists, the  stores analyze historical sales data to pinpoint just how long to hold out before they need to cut a price -- and by just how  much.

Their progress marks a new step in a growing trend toward highly flexible prices -- for everything from mortgages to eBay merchandise. Instead of taking a one-price-fits-all approach, buyers and sellers are increasingly meeting in customized marketplaces transformed by technology.

With exploding competition from discounters and specialty stores,  markdowns are soaring, making them a decisive issue in retailing.  Marked-down goods, which accounted for just 8% of department-store sales three decades ago, have climbed to around 20%, according to the National Retail Federation.

Retailers hate markdowns. Discount an item too late, and stores  are stuck with truckloads of inventory. Too early, and they lose profits as people snap up items thrown on the bargain table  prematurely. Last month, Gap Inc. said its profit margins on June  sales fell well below its internal forecast after it was forced to take deeper-than-expected markdowns on a mountain of merchandise, from Gap T-shirts to Old Navy shorts. And last week, Neiman Marcus Group Inc. cited  steeper-than-planned markdowns for the second time in as many months  in estimating a loss for its fiscal fourth quarter, ended July 28.

Retailing Wild Card

"You can intelligently and consistently predict a lot of other components of your business," says Steve Raish, chief information  officer of J.C. Penney. But "markdowns -- in particular seasonal markdowns -- are one of the least predictable elements."

The technology, still fairly new and untested, requires detailed and accurate sales data to work well. And even if the new software programs can help crack the markdown riddle, they can't solve other  problems, such as poorly chosen merchandise or a weak economy. "This is not the savior to the profitability of the company. This is just  one more tool in the tool chest," says William Podany, the  55-year-old president and chief executive of ShopKo, based in Green Bay, Wis.

End-of-season sales -- held two or three times a year -- first came into vogue in the 1950s, as an easy way to clear inventory and  free up cash. In the 1960s, department stores began adding  back-to-school sales to draw in customers as the summer tailed off.

But as discounters improved their apparel selections over the  years, department stores have found themselves constantly staging  sales to stay competitive. "They don't even name sales anymore, because it would have to be called the 'June 13 sale,' followed by  the 'June 14 sale,' " says Erik Gordon, a marketing professor at the University of Florida's Warrington College of Business.

Much of the attention on markdowns in recent years has been  regulatory, as state attorneys general charged numerous retailers  with deceiving consumers by raising prices and then offering a discount off the inflated price. In some cases, investigators  couldn't find any evidence that the goods had ever been sold at the  so-called "original" price. Just last month, Kmart Corp. responded to a complaint from  the Jewelry Advertising Review Program, a coalition of local Better Business Bureaus, which contended that Kmart was claiming to sell  its jewelry at discounts from "original prices" that weren't  frequently offered. Without admitting wrongdoing, Kmart said it had  changed its jewelry pricing to sell its jewelry at regular prices  for at least 183 days each year.

ShopKo, a chain of 141 stores that likens itself to Target Corp.'s discount stores, began hunting for a technical edge early last year. The 39-year-old  retailer, which also operates Pamida, a 229-store chain of small mostly rural discount stores, was facing a difficult year in 2000,  which ultimately ended with disappointing sales of $3.5 billion and  a $15.8 million loss. It subsequently had to cut jobs, close some stores and put remodeling and expansion plans on hold.

As part of its effort to get a handle on the business, it turned to Spotlight Solutions Inc., a Mason, Ohio, start-up that is one of the leaders in the markdown-software field. Spotlight's software uses mathematical models created by researchers such as Dr. Dale  Achabal, a bespectacled professor who acts as an adviser to the  company and sits on its board.

Trained in marketing and regional economics, Dr. Achabal in 1980 co-founded the Retail Management Institute at Santa Clara  University, a private Jesuit school in California. He currently  directs the institute, which includes the Retail Workbench, a  corporate-sponsored think tank.

Inside the Mission-style building that houses the university's business school, researchers pow-wow with retail executives about  common complaints: managing inventory, measuring advertising effectiveness and improving their supply chains. In one current  project, Santa Clara researchers are analyzing consumer-preference  data to help retailers create the best possible product assortments.

In the early 1990s, retailers complained that they were  overwhelmed by all the sales results bar-code technology was vacuuming up, Dr. Achabal says. "They had more data, but they didn't have better information to make decisions," he says. So he and  others began working on computer programs that would make sense of this mass of data. In 1993, Target installed an early system based on his research.

Impressed with that work, Shopko set up a pilot project last August to test the software, called Markdown Optimizer. Some forecasts from the Spotlight software proved to be dead-on. It correctly predicted that sales of boys' fleece vests would peak in mid-August. So instead of taking its typical markdowns of 10% or 20%  again and again, ShopKo followed the software's recommendation of  taking a single 20% markdown in November, according to Spotlight. It made a 30.2% gross profit margin on the vests during the three-month clearance period -- a huge improvement over the previous year,  Spotlight analysis shows.

By Feb. 1, the end of the pilot project, sales for the roughly 300 products in the test -- everything from sheets to lotion to baby bottles -- were 14% higher than a year earlier. By contrast, ShopKo's overall same-store sales during the fourth quarter were flat. More important, its gross profit margin percentage rose 24%  for the test merchandise. And ShopKo figures it sold 13% more of each product at regular price than it would have in the past.

'Seasonal Demand Curve'

Behind the surprising gains: pricing analysis similar to that  developed by the airlines, which can calculate with great precision  just how many seats to hold open at premium fares for last-minute  passengers and how many to sell ahead of time at lower prices. By  analyzing several years' worth of sales data from similar items, Spotlight's retail software estimates a "seasonal demand curve" for  each new product.

Sometimes resembling a jagged peak, other times a smooth wave, the curve predicts how many units would sell each week at various  prices. For merchandise with short-term appeal -- the bikini, for  example -- sales typically climb for several weeks, spike, then  trail down until the "outdate," or the date a retailer wants to sell out of the item. The software also uses sales history to predict how  sensitive customer demand will be to price changes, what economists call "price elasticity."

To create its new markdown program, ShopKo fed three years' worth  of sales data from each store into Spotlight's computers, which  suggested markdown prices based on how quickly the product category was expected to sell in each of the company's six store groups. The  system uses the data to calibrate a system of mathematical models that incorporate such factors as price elasticity and the number of weeks until the outdate. It then solves equations to determine the  most profitable price cuts for each product.

Software, however, can't predict unexpected events such as a major snowstorm or a surprise sale at a rival store. So it "learns"  during the season, according to the developers of the programs,  adjusting recommendations weekly based on new sales data.

"There will always be things you don't expect," says Scott Friend, president and chief executive of ProfitLogic, a Cambridge, Mass.-based competitor of Spotlight that draws heavily from the  talent pool at Harvard University and the Massachusetts Institute of Technology. But because the software systems start with more precise  forecasts and adjust their recommendations to early sales results, he adds, "on average, we are a lot more accurate" than past practices.

Before Spotlight, ShopKo was drowning in sales data. To determine  which items were stagnating on shelves, store buyers had to sift through stacks of weekly reports with overall sales of each product. The reports also listed inventory levels and how many weeks remained until the outdate. With thousands of different products selling in  more than 100 stores, overwhelmed buyers had to plan most markdowns before each season began. They revised their plans twice a month, marking down an item at the same time across all stores.

That chain-wide approach, common to many retailers, often  sacrifices goods that could have sold at full price in some stores,  and ends up leaving too much merchandise unsold in others. Another  common practice is to chip away at price tags, with lots of small discounts. Dr. Achabal's research concluded that a combination of two markdowns will never be as profitable as a single markdown. Arriving early enough to tempt customers, the first markdown gives the greatest boost to profits, and extra price cuts simply add  profit-eroding labor costs.

Before recruiting Spotlight, ShopKo had tried shifting its markdown dates. At one point, it carried some leftover merchandise  into the next year, a practice it has abandoned. For the most stubborn clearance merchandise, it even offered an extra 25% off. "That created a lot of traffic, but it was terribly hurtful to our gross margin performance," says Mr. Podany, the ShopKo chief executive. "I'm going on my 33rd year in the business, and I've  never mastered this."

Within a few weeks of starting to review forecasts in the pilot  project in October, employees saw surprising recommendations. In its  high-volume "superstores," for example, the average first markdown suggested by the program was 25.7%. In the lowest-volume stores, it was 46.3% -- nearly twice as much. Says Paul Burrows, ShopKo's chief information officer, "We never would have been able to tackle that  manually."

But, right out of the box, the software hit some bumps. On baby  bottles, for example, the program called for an immediate 90%  markdown. Thinking the discount too steep, ShopKo tried it in just a  few stores. There, the bottles sold out far too quickly -- within several days. "People who didn't even have babies were buying baby bottles," jokes Mr. Burrows. ShopKo says that pricing move was an anomaly, caused by a quirk in the selling pattern to which the  system needed more time to adjust. It has since ironed out the  problem by setting a 50% cap on first markdowns.

Cutting Labor Costs

Smarter markdowns also saved labor costs. ShopKo has calculated  that it costs 18 cents to change the price on a single garment tag,  and 24 cents to revise a shelf label. Whenever it lowers prices, clerks scurry through the stores in the early morning, wielding  "price guns" that spit out brightly colored stickers with a new  price. Before using the software, ShopKo frequently marked down  products three or four times. Now, it often does it only once or twice.

Mr. Podany is waiting for a full year of data to pass final  judgment on the new method. But the retailer is moving ahead with plans to use it throughout its stores by this fall, and it even hopes to eventually create a separate markdown schedule for every location.

On many items, the variation can be just pennies. But on more  expensive merchandise, the program may suggest widely different  markdowns in the same town. On a recent day in Green Bay, for  example, one gas grill cost $189.99, or full price, at the ShopKo in the Bay Park Square mall. At a store across the Fox River, just a  20-minute drive away, the grill cost $121.59.

ShopKo is counting on shoppers to gravitate to the most  convenient locations. It doesn't expect many customers to trudge from one store to another to compare markdowns. "I just don't have the time," says Connie Smith, a paper-mill operator leaving a ShopKo store in Menasha, Wis., recently. During the pilot test, ShopKo surveyed its store employees and found that customers didn't seem to notice -- or care about -- different markdowns at nearby stores.

Write to Amy Merrick at amy.merrick@wsj.com

Copyright © 2001 Dow Jones &  Company, Inc. All Rights Reserved.
Copyright and reprint  information.