(FORTUNE Magazine) – Round and round, What comes around goes around. --Ratt, "Round and Round"

Despite the fact that the software industry is careening toward its 50th birthday, it in many ways looks like an industry that has not quite matured--one that is still finding its way in terms of business model and pricing. While Microsoft seeks to move corporate users to a subscription model for Microsoft Office, many application service providers (ASPs) are working frantically in the opposite direction, trying to scrap subscriptions for a model in which cash is collected up-front. How can an industry so old have such schizophrenia about something as simple as a pricing model?

From the beginning, the software industry has had one key characteristic that distinguishes it--variable costs are at or near zero. Economic theory suggests that to maximize profits, you want to pick a price where marginal revenue equals marginal costs. Obviously, when marginal costs are zero, this formula breaks down.

Software companies have tried different ways to address this problem. At the high end, enterprise-software strategists advise high prices to reap the maximum profit and to offset direct sales costs. Companies like Microsoft, though, favor entering the market at a low price with the objective of taking a huge portion of market share. If your R&D dollars are spread across the most customers, no one else can afford to keep up.

Before 1995 most enterprise-software companies followed a pretty consistent pricing strategy: charge as much up-front as possible. Typically, $250,000 seemed about the least you could charge to account for the cost of a direct sales team. On top of that, the customer was asked to pay about 18% per year of the original purchase price in "maintenance" to cover customer support and access to minor upgrades. Then, every three or four years, the vendor would release a "major upgrade," requiring all customers to revisit the big-ticket investment.

In the mid-'90s this model started to show signs of wear. Enterprise-software companies found themselves scrambling to close as many big-ticket deals as possible at the end of each quarter. Once the quarter closed, the company had to start again from square one. Customers caught on and began delaying purchases until the exact end of quarter, extracting better deals from vendors that were particularly eager to close.

This model is not for the faint of heart. Many stressed-out CEOs began to search for a new way to alleviate the uncertainty inherent in the last-minute crunch. About the same time, the rise of the Net gave birth to the ASP--a company that delivers software as a service subscribable over the Web. Analysts loved it: With customers paying an incremental fee each month, the "start from zero" game was gone. Assuming no loss of customers, the revenue from last quarter would be already booked for this quarter--all new sales theoretically represented incremental growth.

Alas, for all the seeming advantages of the subscription model, one key challenge makes it extremely difficult. Let's assume I run a small software company selling enterprise software the old-fashioned way, charging $1 million in year one and 18% maintenance each year after. Sick of my quarter-end ulcers, I switch to an ASP model. The $1.36 million my customer spent over three years is now spread over a 36-month "subscription." But if my salespeople close ten accounts evenly across the first year, I recorded revenue and collected cash flow of only $2.26 million in year one, compared with $10 million in my old model. That is why many ASPs backed off their original pitch and are attempting to sell traditional licenses.

The big problem: capital availability. If a company can break even, then subscriptions are clearly preferred. But the capital needed to grow this model is huge. By pushing customer payments out, the startup is essentially providing vendor financing. When ASPs began to gain buzz, many enterprise-software vendors did the math and called the model "unobtainable."

Ironically, the difficult economy has created a situation in which the customer seems to prefer subscriptions. When capital budgets get cut, people prefer to buy by the drink instead of up-front. Now even the licensed software vendors are entering into financing agreements, with customers paying over time. Once again: schizophrenia.

So what's the best model? Perhaps it's a blend. Recognize revenue on a subscription basis but try to collect as much cash up-front as possible. This will give you a protection from the license model but won't leave you starved for capital. Of course, this model will require enormous patience to reach profitability, but in the long run (forgive me, Mr. Keynes), you will be much better off.

J. WILLIAM GURLEY is a partner with Benchmark Capital, a venture capital firm. Except as noted, neither he nor Benchmark has a financial interest in the companies mentioned. To receive an expanded version of Above the Crowd, visit www.news.com, or to subscribe to the e-mail distribution list, please enter your address at www.benchmark.com/about/bill.html. FEEDBACK: atc@benchmark.com