Dell finds itself in a numbers crunch

The Round Rock, Texas-based PC maker says it inadvertently overestimated the number of machines it shipped during those periods because of a mismatch between its fiscal year and the calendar year. As a result, a discrepancy exists between what the market-share statistics show and what Dell actually shipped to some regions and business segments.

"About a week ago, Dell called and admitted they had inflated their numbers historically," said IDC analyst Roger Kay. Dell attributed the mistake to a "previous team" responsible for the numbers, he said.

Although the total difference appears to be relatively small, it does have an effect. Dell now loses historical market share that it had seemed to have because of the overstatements--but at the same time, its more recent sales growth now looks bigger.

The discrepancy is perhaps widest in the numbers for the European market, where Dell now says it shipped about 200,000 fewer units than reported in fourth-quarter 1999 to third-quarter 2000. Approximately half of the overestimate occurred in the fourth quarter of 1999.

For fourth-quarter 1999, Dell reduced its previously stated European unit shipments by 14 percent, and for first-quarter 2000, it reduced the count by 5 percent, according to market researcher Context. For the second and third quarters of 2000, Dell knocked back shipments 6 percent and 3 percent, respectively.

"This obviously affects the findings of the worldwide market all the way around, and knocked down Dell's market share in some places," said Dataquest analyst Brian Gammage.

Context says that Dell's European market share for fourth-quarter 1999 before the revision was 7.5 percent. With the restatement, the share for that period drops to 6.5 percent. For first-quarter 2000, the market share has been revised downward from 9.5 percent to 9 percent; for the second quarter, the share drops from 9.9 percent to 9.3 percent; and for the third quarter, the share drops from 9.8 percent to 9.5 percent. For all of calendar 2000, Dell's share drops from 8.9 percent to 8.8 percent.

Growth gains
Although revising sales figures downward would seem to be a negative, it could help Dell in some respects. In Europe, for instance, half of the overestimated shipments were attributed to 1999. By removing the phantom units, sales for 1999 sink--but by the same token, the growth rate in 2000 for Europe rises substantially.

Dataquest, for instance, had Dell's European unit shipments down 3 percent year over year in the fourth quarter of 2000. If the "correct" numbers are used, growth in shipments stands to show an increase of more than 28 percent. Dataquest, however, is not changing its historical data.

Context, meanwhile, before the adjustment put Dell's fourth-quarter 2000 European sales growth at 6.4 percent, or about 2 percent below the overall European market. But by removing 100,000 units from the year-earlier quarter's shipments, Dell's growth jumps to more than 31 percent based on the more favorable comparison.

"The restatement of the numbers means Dell's growth figures look better than they actually were," said Jeremy Davies, senior partner at Context.

Dell spokesman Jim Mazzola acknowledged the benefits of the revision. "Year over year, yes, it would certainly boost our growth," he said, adding that "we're not boosting our numbers or anything like that."

The problem emerged because of the discrepancy between when the market-analysis companies need sales data and when Dell closes its books. Dell's fiscal year ends in January, and its financial quarters end a month after the calendar quarters. The analysts, on the other hand, typically request shipment data that reflects the calendar quarters. Dell has been providing estimates for the fiscal quarter, and since the data is being requested a month before its end, Dell has estimated the sales on the coming month. Those estimations, in retrospect, were a tad high.

"We count by calendar quarter, but Dell has actually been communicating based on their fiscal quarter. This we didn't know," Gammage explained. "Because they were doing it in that format, they were giving two months of real data and one month of projected data."

Though Dell has spoken up about the reckoning problem, Kay said there is uncertainty whether the problem has been or will be fixed. "Dell said it was their investor relations people that insisted on reporting by fiscal quarter," Kay said. "My guess is that hasn't changed."

An unusual request
Mazzola defended the company's position, insisting that "restating is not uncommon."

But some analyst firms bristled at the request. Context has decided to publish two sets of numbers for the rest of year, one based on the original figures and another on the revisions.

"From a research point of view, it was very difficult for us," Davies said. "So we have to go back and start looking at the numbers in light of what they said and figure out how we're going to deal with it in the future."

Gammage considered Dell's request highly unusual. "It's almost unique to restate so far back," he said.

Dataquest chose not to change historical data from fourth-quarter 1999 to third-quarter 2000, but is lowering the estimate for fourth-quarter 2000. Dell didn't ask Dataquest to examine the most recent numbers, but the research company did substitute calendar-quarter for fiscal-quarter numbers.

"We don't restate historical numbers," Gammage said. "If you take something a year back, you change the whole way the market looks, and that change would have had a big effect a year back."

Nonetheless, by altering its most recent quarterly numbers for Europe, Dataquest gives Dell a black eye and shows the company shrinking in Europe because the Dataquest is comparing the "old" inflated numbers with smaller current figure.

"Especially in Europe we have a strong December and a strong January, but we know that October is quite weak," he said. "Based on that, to Dell's surprise, we revised the fourth-quarter 2000 figures downward. Obviously, this isn't what Dell wanted us to report to the market."

IDC showed the greatest willingness to revise numbers based on Dell's request. "We revise figures all the time," Kay said. "If someone presents us with corrected information, we can't ignore that. I call it 'retrofitting.'"

Mazzola insisted that there is nothing wrong with the reporting because Dell presents its own data to Wall Street.

"We've got our actual data, so we can report that in our data," he said. "Whether the (analyst) firms have it accurately as we reported or not is up to them."

But market analysts point out the third-party data is essential to Dell and other manufacturers. They use the data for marketing purposes and to show Wall Street their growth and position against competitors.

Kay also noted that not all the adjustments are downward. In the case of the U.S. education market, Dell claims that it shipped more systems than previously thought. IDC plans to revise past figures, putting Dell ahead of Apple Computer at least one quarter earlier than expected.

"Dell wanted to restate two quarters of data and essentially prevent Apple from claiming it had been No. 1 in education in 2000," Kay said.