As Silicon Valley start-ups go, it's hard to find a simile for Metaphor -- it's just not like other fledgling computer companies.

Metaphor Computer Systems is "breaking all the rules on how to succeed in this business," industry analyst Esther Dyson said. "It's like they're saying 'yeah, but we think we can carry it off because we're so good.' "

Metaphor's unspoken arrogance is matched by one of the more unorthodox approaches to a turbulent office-automation market crammed with competition and dominated by an increasingly aggressive International Business Machines Corp.

Instead of selling general-purpose computer systems that address a broad spectrum of applications, from air-traffic control to accounting, Metaphor is custom-tailored to a group of users -- individuals who use information, who make reports and analyses and studies.

The company is targeting the Fortune 500 marketing managers, the financial specialists and the market analysts who need to examine, analyze and package data into useful information for top management.

It offers them an extremely powerful -- and expensive -- system of high-tech work stations connected to a company's big computers. The system is designed to give people a sharper focus on the data that's constantly pouring into the corporate mainframes.

Metaphor believes that its brand of customized computing, which allows it to skirt direct competition with companies such as IBM and Apple Computer Inc., will enable it to carve out a comfortable $150-million-a-year niche within three years.

"We have this magazine ad -- it's an allegorical ad . . . ," Metaphor Chairman David E. Liddle said. "It shows a manicured corporate lawn somewhere in Connecticut. On one page, a big crowd of happy, smiling people are standing on the lawn and on the other is a smaller group of unhappy-looking people on the right.

"The caption over the crowd reads, "All of these people's jobs have been improved by the IBM Personal Computer -- but the 250 people on the right can't be helped."

Those are the ones "whose offices are filled with printouts marked by yellow highlighter," Liddle said. They're the information users who fall between the cracks of corporate computing: On one hand, their corporate data-processing department isn't equipped to cater to their data demands and, on the other, personal computers just aren't powerful enough to access the needed data or model it into useful forms.

Of course, some companies are trying to plug that gap by linking personal computers to the corporate computers but, for the most part, that's been an ad hoc process. According to Donald Massaro, Metaphor's president, what the firm is pushing is "a system -- not a stand-alone" solution to the problem.

Metaphor borrows heavily from technology pioneered at the Xerox Palo Alto Research Center -- much in the way that Apple Computer's Macintosh is a Xerox research spin-off.

Like Macintosh, Metaphor takes a push-button approach to data processing, rather than one that relies on text displays. The high-resolution screen displays a set of icons (Metaphor calls them metaphors) that illustrate the task to be performed.

By using a interface device called a mouse, the user can point to the desired function on the screen and then touch a key on a five-function keypad to specify the activity. For example, a Metaphor user can examine a table of numbers and then, by pointing to the appropriate rows and columns, convert it into a pie chart or a bar chart. Another tap of the function keys will enlarge or shrink the chart. Another few clicks, and the chart can be inserted into a report.

Relying heavily on this "point and click" approach, Metaphor offers a "keyboardless" way for users to assemble and analyze disparate models of data.

Basically, think of the Metaphor screen as a lens to view and magnify corporate data from a variety of angles. "It does change the way that I, as a manager, can get things done," said Kerr Curtis, director of Bank of America's Retail Banking Division. "I can get quicker turnaround on projects. It's a lot more significant than adding a few personal computers, but it's not what I would describe as revolutionary."

However, Curtis said that "we're high on it" and implies that the bank is seriously considering a major investment in the system.

Metaphor's unique approach presents it with a dual marketing challenge: The first is to sell the concept of Metaphor's database and interface abilities to the end user, and the second is to convince the corporate data-processing department to invest the resources to link Metaphor to its mainframes.

"They've got to convince the data-processing department that they're helping them rather than usurping them," industry analyst Dyson said.

Another key aspect is that Metaphor isn't a discretionary corporate purchase like a dozen or so personal computers. It is a big-budget capital expenditure that may have strategic implications for a company's marketing and data-analysis approach. The average Metaphor installation costs $152,000 and a minimum configuration of only eight work stations will set a company back $64,000.

Metaphor, of course, considers its system worth every dime.

"This is a system that's designed with ears," said Casey Branscomb, Metaphor's 29-year-old vice president of marketing, alluding to the more than 250 inverviews that Metaphor conducted at 40 companies to find out what users wanted from personal computing.

Branscomb -- the daughter of IBM's chief scientist, Lewis Branscomb -- stresses that Metaphor is marketing "solutions to business problems rather than a technology" to its potential customers.

The company has taken the unusual step of creating its own direct sales force to call on Fortune 500 companies to pitch the system. Each sales team consists of a both a technical person who can work with a company's data-processing department and an "applications analyst" -- often someone who worked in the target industry -- to explain to the end users how Metaphor's interface can be used to add value to corporate data.

Metaphor concedes that direct selling is expensive, adding to the company's already high-cost structure. But Metaphor's management is convinced that, by virtue of its huge research, development and production investment -- as well as its targeted sales approach -- it is erecting a powerful barrier of entry to competitors.

"When Metaphor gets done spending this money, it will have a product and direct sales force that is not easily duplicable. But they have to execute efficiently and crisply now," said William Hambrecht of Hambrecht & Quist, the San Francisco-based venture capital firm that has poured more than $8 million into Metaphor.

While Metaphor may succeed in blocking out entrepreneurial challenges, the shadow of IBM looms large over its market.

Massaro said that economies of scale make Metaphor's niche too small for IBM's attention: "They're the Toys R Us of the data-processing department -- they go for volume."

Yet, as dozens of other start-ups have shown, IBM has become far more aggressive in the pursuit of markets and could make future growth and profitability very difficult for a Metaphor if it develops a rival bridge between its personal computers and mainframes.

Conversely, Metaphor might find itself the computer version of the Concorde -- a technology that does things sleeker and faster than before, but at such a cost that few firms are interested in that big an investment.

Those problems should not obscure the fact that Metaphor offers a view of what the future of office computing may look like: the integration of work station, network and mainframe at the touch of a few buttons. It is an approach that probably will be imitated by other companies seeking solutions to office automation.