2000 ANNUAL FINANCIAL RATIOS.

It pays to be big. That's the basic lesson of our latest Soft-letter survey of topline financial benchmarks, which this year shows that the industry's largest public PC software companiesn (those with revenues over $100 million) have taken a strong lead in almost every major aspect of business performance. Marketing, operations, product development--the big guys now run their companies a good deal more efficiently than their smaller counterparts, and the inevitable result is that big companies as a group are also the industry leaders in profitability.

Moreover, the financial gap seems to be widening. Although there are large companies that lose money and small companies that are highly profitable, median operating profit for big software companies rose from 5.1% of sales in 1998 to 7.3% in 1999, while more than half of small and mid-sized companies lost money during the same period. Predictably, a good many of these troubled companies have already found acquisition partners or are shifting their focus to the Web, to videogame platforms, or to enterprise market segments.

At the same time, this year's survey data is a reminder that it takes more than size to succeed in the software business. Software companies often turn from cash cows to financial disaster zones (and back) in just a few fiscal quarters. Last year, for example, Network Associates saw operating profits plunge from 12% to a -20% loss, while Macromedia jumped to 15% profitability after years of hefty losses. Other good- sized companies--Inprise, Intuit, Wall Data, GT Interactive--have looked for the magic "economies of scale" in software and have found little more than red ink.

This kind of extreme volatility, though rare in other industries, has always been typical of technology companies. And because the numbers bounce around so much, it's fair to ask whether statistical benchmarks like ours are actually meaningful for analyzing performance and setting budget targets. Do the numbers measure anything real, or are they statistically random? And what's being measured when the data comes from companies that sell products as diverse as games, developer tools, and accounting software?

In fact, we've found that the numbers are surprisingly consistent, even among diverse companies. That's probably because the ratios we report sum up the collective experience of literally thousands of people who manage software companies. Years of collective experience suggests, for instance, that most software companies will need to spend between 35% and 50% of their revenues on marketing. Similarly, a reasonable level of R&D investment seems to be about 20% of sales, and COGS (Cost of Goods Sold) spending is likely to add another 20%. There are certainly successful software companies that end up with different ratios, but almost always because of unique market conditions (such as a near- monopoly in a product category) or unusual business models.

Moreover, these ratios reflect the fact that software companies, regardless of what markets they serve, tend to pay roughly comparable salaries to recruit programmers and sales reps, to buy magazine ad pages and direct mail, or to rent office space, borrow money, or pay the phone bill. How much profit a company extracts from these resources is a matter of management performance, but the underlying costs are similar for almost everyone--and the spending ratios for these costs end up being similar as well.