Fool.com: Burning the Barron's (News) August 2, 1999

ONE FOOL'S OPINION

## Burning the Barron's Amazon.com Under Fire -- Again

August 2, 1999

It's unfortunate for readers of Barron's that the old-fashioned "doom sells" magazine is published every Saturday. Not being a frequent reader, I can just imagine how the gloomy copy can dampen one's weekend activities. This weekend was no different. Nestled between the usual array of self-important downbeat columns, yet another negative Amazon.com (Nasdaq: AMZN) article was given bright lights on Barron's dark pages. The article took some 1,000 words to repeat several times that revenue per customer at Amazon declined last quarter and has been declining since March of 1997, with the exception of December quarters.

Revenue per account is typically -- and predictably -- highest in the fourth quarter of each year, as it was in both 1997 and 1998. In other quarters, it declines per account. The author of the Barron's column pointed to the declining revenue per account figure as a serious caution flag, a flag that blows strongly enough to merit, in his view, the question of whether or not Amazon's business model will even work.

While it is entertaining to think that this one mathematical calculation alone could render a business model inoperable, the reality is that Amazon's business model has already created \$16 billion in market value, compared to the \$4.4 billion created at Barron's parent company, Dow Jones (NYSE: DJ). This nitpick aside, the notion that a revenue per account figure could be Amazon's death knell is overblown.

In the second quarter of 1999, revenue per customer did decline to \$29, one dollar below the average estimate. Meanwhile, customer accounts increased 2.3 million to 10.7 million. That was 400,000 more than the typical estimate. The answer to the issue of revenue per account raised by Barron's lies, in the end, in the number of total customer accounts.

Amazon ended 1997 with 1.5 million customer accounts and an average of \$44 in quarterly revenue per account. In 1998, it had 6.2 million accounts and \$37 in average quarterly revenue per account. Now at 10.7 million accounts, so far in 1999 the company has averaged \$32 in revenue per account. The long-term estimate -- for 2000, 2001 and beyond -- calls for \$29 to \$31 in quarterly revenue per account. Amazon is above that now (at \$32) and the fourth quarter has always seen a jump in the past. Revenue of about \$33 per account is anticipated in the fourth quarter, following quarter three revenue of about \$27 to \$28 per account.

The more accounts that Amazon has, the more likely it is that many of the accounts will not buy anything from Amazon during a particular quarter. Of course, during Amazon's first year as a public company, when it had only 1.5 million primarily new customers, most of those customers shopped. Thus, the revenue per account was high. In 1998, revenue per account at Amazon declined as the number of accounts grew, partially because not everyone shopped the site.

Over 60% of customers returned to make a purchase in any particular quarter in 1998, but the nearly 40% who didn't shop during any particular quarter were still counted in the total account mix, and thus they served to lower the average revenue per account. What is most important is how many total customers Amazon has, however, and how many customers return over the year rather than every single quarter. A vast majority of customers (at least two-thirds) still return to the site at least once a year to shop. So, even though revenue per account declines as the customer base expands and 30% of customers don't return, it's still better to have more total customers (even if some are inactive) and dilute this "per account" number than to have fewer customers period.

Barron's seems to disagree.

Barron's is effectively arguing that having more customers is worse than having fewer. How are they arguing this? Because they argue that declining revenue per account since 1997 is a horrible thing in and of itself. They apparently do not realize that a continually growing customer base all but guarantees that average quarterly revenue per account will decline. This is almost a certainty of mathematics. And yet Barron's is arguing against it.

In the end, average quarterly revenue per account of about \$29 to \$31 is anticipated long-term, and so far this year Amazon is at \$32 per account. Revenue growth is expected to accelerate again in quarter three and especially in quarter four. This brings up the final point made in Barron's. The author reminded everyone that sequential revenue growth is slowing. He concedes that as the revenue base balloons, sequential growth slows on a percentage basis. We all understand that. However, Barron's doesn't admit that the absolute sales dollar total at Amazon is hitting a new record every quarter. Merrill Lynch anticipates year 2000 revenue of \$2.3 billion, up from \$1.45 billion in 1999. This year, \$1.45 billion in sales would more than double last year's \$610 million.

Amazon has been a holding in the Rule Breaker Portfolio since September 1997, when it was purchased at \$6.58 per share.

By Jeff Fischer (TMF Jeff)