Steady sales growth helped Mercury Interactive ( MERQ) navigate the tech bust better than most of its peers. But an increasing emphasis on subscription sales has prompted consternation about how to value the application software maker.

The controversy over Mercury only seems to be rising: Mercury Interactive recently attracted the rare sell rating from one analyst and some investors claim the stock is too expensive, even as others rush to the firm's defense.

Although the stock is off more than 21% from its 52-week high as of Thursday's close, it's still too pricey, says Rich Parower, co-manager of the ( SGTRX) Seligman Global Technology fund. "We're GARP-y and we mean it," he quipped, referring to the growth-at-a-reasonable-price style of investing.

The problem for many investors is Mercury Interactive trades at nearly 40 times the 2004 consensus earnings estimate gathered by Thomson First Call vs. the 30-times benchmark for typical software companies.

But boosters of Mercury, whose products are used to test and manage application software, say the stock shouldn't be valued as a traditional software company. About three years ago, Mercury Interactive changed its business model and began selling its application performance management products as a multiyear subscription, a break from industry norms.

Traditionally, software companies sell software as a perpetual license in which they immediately recognize revenue from upfront sales.

Under the subscription model, however, the bulk of a sale is first recognized on a company's balance sheet as deferred revenue and then moves to the income statement over the contract's life. Consequently, Mercury's revenue -- and ultimately net income -- appear lower than if the company recognized all of the revenue upfront.

Lower earnings, in turn, make the company's stock look more expensive than its peers based on a price-to-earnings ratio.

"As more of the company's sales 'go to the balance sheet' in the form of subscription bookings, the income statement alone does not fully measure the progress the company makes in a given quarter," Pacific Crest analyst Rich Petersen wrote when he upgraded Mercury to a buy last week. (His firm hasn't done banking with Mercury.)

Goldman Sachs analyst Sarah Friar echoed that sentiment. In a recent note, she observed that Mercury's multiple is far less worrisome -- and even low -- when based on free cash flow. Based on Thursday's closing price, Mercury is trading at nearly 21 times Friar's 2004 estimate for free cash flow vs. her view that 26 to 28 is a reasonable cash flow multiple for software companies.

Another way to compare Mercury to its peers is by backing out deferred subscription revenue to calculate revenue and earnings as if all of Mercury's subscription deals were perpetual licenses, Friar wrote. She found 2004 earnings would range from $1.70 to $1.80 a share -- vs. her $1.13 estimate under the subscription model -- resulting in a more reasonable P/E of about 24 to 25. (She has an outperform rating on the stock and Goldman has done banking with Mercury.)

Impressively, Mercury's deferred revenue line has grown sequentially and year over year every quarter since at least the beginning of 2002. In each quarter of 2003, deferred revenue grew between 147% and 194% year over year and between 20% and 46% sequentially. (By contrast, Microsoft ( MSFT) recently suffered a steep decline in deferred revenue as some customers have resisted its move to a subscription revenue model.)

"With a subscription revenue business you have a great deal of visibility," said David DuShene, a software analyst at Firsthand Funds, whose ( GTFQX) Global Technology fund includes Mercury among its top 10 holdings. That's because you can calculate roughly how much deferred revenue from the balance sheet is waiting to flow to the income statement, reducing the dependence on new license deals to some extent, he said.

"In my mind, I'd pay up for that," DuShene said. "I think that deserves a premium."

Is Mercury Pulling a Fast One?

However, Mercury may have created additional confusion when it changed the language used to describe subscription revenue to "term licenses" from "subscription fees" in the third quarter. Oppenheimer analyst Sanjiv Hingorani questioned why the company switched terms earlier this month while initiating coverage of Mercury with a sell rating.

Company officials declined to comment on this change and also turned down a request for an interview, citing a quiet period.

Hingorani argued that license fees -- from both perpetual and subscription licenses -- are not growing as fast as many investors believe. That's because some portion of the company's subscription fees actually include maintenance fees that customers pay for support services, he argued.

"We believe that as investors realize that the company's 'true' license revenues may not be as robust as they appear to be, there is a likelihood that there may be some compression in the P/E multiple accorded to the shares of Mercury Interactive," Hingorani wrote.

The company should disclose what portion of its subscription fees are actually maintenance fees, he said. In its 10-K, Mercury maintains it can't determine the value of the license fee separately from maintenance because the two are bundled together in subscription contracts.

Assuming that 20% of subscription fees should be categorized as maintenance revenue -- the typical breakdown for perpetual software licenses -- Hingorani found Mercury's overall license revenue grew 2% to 3% less per quarter in 2003 than the company reported. Last year, the firm's quarterly license revenue growth ranged from 14% to 27% on a year-over-year basis.

But Mercury supporters contend Hingorani's analysis is flawed. Kevin Merritt, an analyst with Fiduciary Trust Co. International, believes Hingorani oversimplified the matter by assuming 20% of subscription revenue should go to maintenance revenue.

That proportion may be the general rule of thumb in a perpetual software license, in which a maintenance agreement entitles a customer to software updates as well as support. But under a term license or subscription, maintenance includes primarily customer support, and after the term license expires in one to three years, a customer has to go out and sign up for another subscription, Merritt said.

Merritt, whose firm holds Mercury shares, believes it would be more accurate to allocate 5% to 8% of a license to maintenance.

In a follow-up interview, Hingorani defended his calculation, noting that maintenance included in subscription contracts does, in fact, offer updates. He cited the company's 10-K, which says "subscription revenue, including managed service revenue, represents license fees to use one or more software products, and to receive maintenance support (such as hotline support and updates)."

Ultimately, the question of how to properly allocate maintenance revenue is just another element in the simmering debate over Mercury's worth.

More from Technology

Experts Break Down GDPR Risks for Investors

Experts Break Down GDPR Risks for Investors

60 Seconds: What the Heck is GDPR?

60 Seconds: What the Heck is GDPR?

Netflix Ready to Surpass Disney as America's Most Valuable Media Company

Netflix Ready to Surpass Disney as America's Most Valuable Media Company

Musk's Tweets About Unions Draw Fire From United Auto Workers

Musk's Tweets About Unions Draw Fire From United Auto Workers

PayPal Wants to Consolidate the World of Rewards Points

PayPal Wants to Consolidate the World of Rewards Points