(FORTUNE Magazine) – NOW THAT the global marketplace has reached adolescence, it seems almost everyone is under the covers with everyone else. IBM alone has joined in over 400 strategic alliances with various companies in the U.S. and abroad. In July, Big Blue teamed with Siemens and Toshiba to develop a costly new generation of DRAM computer chips, and that was just one of a score of recent announcements by AT&T, GE, Merck, Time Warner (the parent of FORTUNE's publisher), Matsushita, and Fujitsu, among others. Alliances have become an integral part of contemporary strategic thinking. McKinsey's David Ernst and Joel Bleeke, co-authors of the forthcoming book Collaborating to Compete, say the rate of joint venture formation between U.S. companies and international partners has been growing by 27% annually since 1985. Are these partnerships achieving the goals set out for them? In many cases the answer is no: Roughly one-third of the 49 alliances tracked by McKinsey were flops, failing to live up to the parents' expectations. But through such painful lessons, managers are growing smarter about what to expect from these linkups and how to craft a winning alliance. As a result, today's corporate partners are less interested in short-term ventures designed to save a few & dollars and more focused on long-term alliances where gains can be harvested for years to come. Says Ram Charan, a consultant whose clients include GE, Du Pont, and Citicorp: ''Saving cash alone is not enough to justify an alliance. The reason to do an alliance is to get access to a new market or a special expertise, or to beat others to market. If an alliance won't achieve any of those things, don't do it.'' Like newlyweds, newcomers to corporate alliances are often surprised by the demands placed upon them. The difficulty lies not only in the management of the business but also in fuzzy areas, such as the personal relationships between managers from different corporate cultures. When conflicts erupt, they are typically much harder to resolve than in conventional companies, where a top executive can usually end an internal dispute by fiat. The only balm that seems to work is time-consuming contacts between senior managers at the participating companies. The potential damage from these conflicts can be lessened, managers are learning, through agreements that are carefully crafted at the outset. Merck senior vice president Francis H. Spiegel Jr. has put together major ventures between the pharmaceutical powerhouse and Johnson & Johnson, Du Pont, and AB Astra of Sweden. Says he: ''The deal structure has to be very creative. You can hammer out a megamerger deal in 60 days, but these joint ventures can easily take a year to negotiate.'' The best deals, Spiegel says, are flexible enough to change with the business. In the Astra alliance, Merck started out simply with U.S. rights to its partner's new drugs. For the next phase, starting in 1993, Merck is setting up a new corporation to handle the partnership's $500-million-a-year business and will sell half the equity to Astra. With the Du Pont deal, the partners each got half of the equity, but to reward performance they agreed to give most of the profits to whichever company kicks in the best products. Corning, the $3-billion-a-year glass and ceramics maker, is renowned for making partnerships work. Dow Corning, its joint venture with Dow Chemical, is listed in the FORTUNE 500. Among Corning's other bedfellows are Siemens, Germany's electronics conglomerate, and Vitro, Mexico's biggest glassmaker. Alliances are so central to Corning's strategy that the corporation now defines itself as a ''network of organizations.'' Its success at corporate linkups comes from working with a very long term perspective. ''We're looking only for lifetime associations, because you have to invest an enormous amount of energy to make a partnership work,'' says vice chairman Van Campbell. ''You not only have to deal with the business; you also constantly have to deal with the relationship you have with the partner -- nurturing it and maintaining high-level contacts, so that when you deal with items of substance you will be dealing with friends, people you understand and respect.'' A partnership that is going to last only five to seven years, Campbell adds, simply doesn't warrant that kind of investment.

Alliances are also more likely to succeed when the partners behave as equals even when they're not. Deals needn't be structured on a strict fifty-fifty basis; what's essential is the spirit of cooperation and attention to the ongoing advantage of both partners. If one feels disadvantaged, even slightly, odds are the deal won't last. One such failure was TRW's alliance with Fujitsu. Hoping to get more out of its U.S. distribution organization, TRW in 1980 signed on to sell some Fujitsu products, including point-of-sale terminals and automated-teller machines. But TRW's salesmen couldn't move the goods and blamed the Japanese partner for failing to rapidly adapt them to U.S. needs. The folks at Fujitsu weren't happy either, complaining that they weren't having sufficient input on decisions. In 1983 the union broke up, with Fujitsu buying out TRW's interest. Acting as equals does not necessarily mean sharing decision-making. Whether because of distrust or a genuine desire to share, partners often try to run their joint ventures together. The result is usually management gridlock. McKinsey's study and the advice of alliance veterans suggest two alternative strategies: Either give one partner sole authority to run the venture or set the alliance up as a completely autonomous operation answerable only to its own board. The rationale for giving one partner operating control is typically based on that partner's superior knowledge or skill. Merck will call the shots in the early stages of the Astra joint venture because Merck knows the U.S. drug market. Since Ford bought 25% of Mazda in 1979, it has formed a web of cooperative projects with the Japanese carmaker. In the U.S. market both companies sell similar cars made at the same plant. Mazda markets Ford products under its own nameplate and also engineers Ford's small cars such as the Mercury Tracer, while Ford takes care of light trucks such as its Ranger. Autonomy is emerging as an important ingredient for success in alliances because it allows a new venture to adapt to its particular market instead of aping the practices of its parents. Says Peter Schavoir, IBM's director of strategy: ''IBM will expect to transplant its culture into the new company, but you can't do that. A startup doesn't need all our planning, control, and measurement systems. You're better off starting from scratch, designing new systems for the new company.'' Perhaps the biggest stumbling block to the success of alliances is the lack of trust. ''When people mistrust each other, greed and selfishness creep in,'' says Richard Miller, CEO of Wang, an alliance partner with IBM. But trust can't be written into a contract; it has to come voluntarily. As McKinsey's Ernst and Bleeke see it, it's in these subtleties of relationship building that American companies come up short. Says Ernst: ''We Americans are used to acquisition and control. That's our mind-set. The lesson for U.S. companies is that it's necessary to have the mind-set of collaboration.'' FORD'S PRESIDENT, Philip Benton Jr., thinks the need for trust is yet another reason that American companies have to lengthen their time horizon on alliances: ''It's absolutely essential to start slowly to build trust,'' he insists. ''The first time two companies work together, the chances of succeeding are very slight. But once you find ways to work together, all sorts of opportunities come up -- and your likelihood of severing that relationship and starting the whole process over again becomes remote.'' Benton argues that because Ford and Mazda have established a good relationship, the two companies should link up even on projects where one partner may be less than optimally qualified. As he explains, ''Learning to work together is so hard, it would be foolish to throw all that away just because you see another girl across the street.''

In business relationships, economic value helps create trust. Just as no amount of New Age executive warmheartedness can salvage a crummy business, winning together in the marketplace is a great way to help partners overcome their differences. Microsoft has maintained and even strengthened its alliance with Apple Computer, supplying applications software for Apple's Macintosh, even though the two companies compete directly in some areas and Apple is alleging in court that Microsoft's Windows software constitutes theft of Apple's intellectual property. ''I feel fine about that,'' says Steven Ballmer, senior vice president of Microsoft. ''This is business. We're not allied with Apple out of love.'' Microsoft's software has helped Macintosh gain acceptance in the business community. Microsoft is making out too: Sales of its applications software for Macintosh totaled $265 million in fiscal 1992, up 65% from 1991. The most important lesson for managers heading down the aisle is to remain flexible. McKinsey's data show that most successful alliances change dramatically during their first few years. Markets keep evolving, and so do the venture's parents. William Clapper, head of planning at GE Aircraft Engines, points to one common source of stress: ''If you're improving, your partner has to improve too.'' When alliances can't adapt, most fail. One of the classic failures was the long linkup between Microsoft and IBM. An important reason for Microsoft's fast rise from almost nothing to a market value of $21 billion was its relationship with Big Blue, starting with a contract to supply the basic software that makes IBM-type personal computers work. The IBM relationship was so crucial to Microsoft that Ballmer says he devoted ''30% to 50% of my brain cycles'' to it for a decade. Then a dispute about how to design the next generation of software broke up the partnership, creating painful uncertainties for their customers. Explains Ballmer: ''It's much harder to change direction and do what customers want in the context of an alliance.'' Joint efforts need clearly articulated ground rules and goals, but they need plenty of elbow room too. Any agreement should detail procedures for revaluing partners' shares and rethinking the partnership's direction. No rules can guarantee a happy marriage, but those pointers can do a lot to bring down the rate of divorce.