(Business 2.0) – Just days after announcing the merger of Hewlett-Packard and Compaq in the fall of 2001, Carly Fiorina flew to Houston for a gut-check summit with her counterpart, Compaq CEO Mike Capellas--and to make perhaps the most important introduction of her career. At Fiorina's side was Webb McKinney, the 33-year HP veteran who was considered the company's organizational mastermind. With Capellas was his young chief financial officer, Jeff Clarke, a numbers whiz whom Capellas admired as a bare-knuckles negotiator.

The two men had never met--"I didn't know Jeff from a hole in the wall," McKinney recently recalled--but they might have suspected that they were about to become inseparable. Fiorina and Capellas had picked them to tackle one of the toughest jobs imaginable: moving the largest technology merger in history from paper and promises into a functioning organism, in the midst of the worst technology slump in two decades. In the process, McKinney and Clarke would lead the redeployment of a combined 145,000 workers in 160 countries. They would be responsible for untangling 163 overlapping product lines--from Intel-based home computers to Unix workstations to handhelds--and firing more than 15,000 employees to meet the $2.5 billion in cost reductions that Fiorina was promising investors.

That Webb McKinney and Jeff Clarke (or Weff, as HP workers began to call the pair) hit those cost targets by last December, 18 months ahead of schedule, and are on track to save another $500 million this year is impressive enough. But the savings achievements aren't the reason Tom Ridge, for example, recently sought HP's advice on how to merge dozens of far-flung government agencies into his new Department of Homeland Security. What interests Ridge and business merger planners alike is the formula HP put in place to finish the task.

Corporate mergers have, at best, A spotty history of success, and technology marriages have always been the hardest to get right. One reason is that tech firms are far more dependent on know-how than, say, a supermarket chain, and defections--a temptation in any merger--can instantly sap a company's value. Thirteen months after 3Com bought Palm Computing in 1997, for example, Palm's executive brain trust, founders Jeff Hawkins and Donna Dubinsky, left to start rival PDA maker Handspring. 3Com spun Palm out a year later, but after a blockbuster IPO, the company began a long decline.

Keeping pace with new products during a merger also plagues tech firms, where product cycles pass in months, not years. After acquiring supercomputer maker Cray for $767 million in 1996, SGI struggled to introduce high-end servers based on Cray's vaunted technology. It watched as rival Sun Microsystems passed it by with faster machines--using Cray designs it had bought from SGI a year earlier.

Banging out new products at a speedy clip, though, is a cinch compared with what plagues mergers most of all--the inability to make fast decisions. Few knew that better than Clarke, who had a ringside seat during Compaq's $9.6 billion acquisition of Digital Equipment in 1998. By the time the deal closed later that year, merger executives had failed to trim redundant product lines, and customers were left guessing about the fate of Digital products, such as its high-end Alpha servers. Amid the confusion, Dell took just 15 months to slip past Compaq as the leader in domestic PC sales. "We screwed up," Compaq founder Ben Rosen said. "We learned you have to make a lot of correct decisions quickly."

Step 1: The Decision Factory

That's exactly what McKinney had in mind when he sketched out the structure of HP's integration team, one designed to have the right alchemy with (and constant access to) top management. Days after the Houston meeting, he and Clarke began recruiting managers in equal numbers--Clarke rounded up Compaq talent, and McKinney lined up their HP matches. In just six months, the integration group, called the "clean team," would grow much faster than any startup. Within weeks of the merger's announcement, the team had 500 members; by March 2002, more than 900. Even after the merger closed in April 2002, it kept growing, peaking at more than 1,000 full-time employees.

By establishing such a huge body of managers and reassuring them that their jobs would be safe even if the merger failed, Clarke and McKinney were able to coax them to share in confidence everything they knew. It also kept most of them motivated to stay--another critical benchmark.

Step 2: "Adopt-and-Go"

Just as important was creating an assembly line for decision-making. In most mergers the power to say who goes and who stays is weighted in favor of the acquirer. HP weighted it equally. The strategy, called adopt-and-go, was to get cross-company pairs of managers to meet daily to determine whose products had the most market share, the better brand, and so on. Then, at weekly presentations with McKinney and Clarke, managers had to offer up one for elimination. In four months, this game of managerial Survivor yielded a road map for product lines and helped close redundant warehouses and factories, ultimately saving $500 million in procurement costs. In the end, many Compaq products beat out HP's (see chart, above). But when it came to choosing a brand, HP was the clear winner. On business machines, HP will soon phase out the Compaq name.

Adopt-and-go went beyond generating a product lineup. It helped HP get through the most painful part of integration planning: generating 18,000 pink slips. The process led to hard feelings in certain quarters. One of them was HP's sales team, which surrendered key positions to Compaq's aggressive and well-regarded sales managers. "The perception was that Compaq people got all the big jobs," says Mike Cox, executive vice president for sales and delivery at IT services firm Logical, who left HP in early 2001. But those decisions, Clarke contends, went by the book. Compaq veterans earned the top spots in U.S. sales by delivering better performance.

In June 2002, Clarke complained to Fiorina that management job cuts (not among her direct reports, but in the layers below) were falling short. Managers who hadn't been named to posts were still collecting checks. Fiorina insisted that Clarke get the process back on track, and within two weeks, hundreds of managers--including many HP career veterans--were gone.

Step 3: Feed the Fast Track

From the start, McKinney set up the clean team to ensure that the process wouldn't slow product launches. Once adopt-and-go had run its course, new-product managers got the resources they needed to keep moving. Last November, HP launched its Tablet PC and began selling the $1,399 Media Center PC before either Dell or Sony could respond. The company has also kept feeding its profit engine--the printer business--with dozens of new product launches, keeping the pressure on Canon, Lexmark, and the latest and most dangerous newcomer, Dell. "When you start doing [merger planning], you run the risk of taking your eyes off the business," says Mark Sirower, a corporate merger consultant. "That kills a lot of deals."

Step 4: Enforce Cost Deadlines

If McKinney was seen as CEO of the integration team, Clarke served as his tough-minded CFO. "If there were issues that the teams couldn't resolve," McKinney explains, "Jeff and I would jump in." If those two couldn't resolve the impasse, they'd pass it to a committee chaired by Fiorina.

One such issue was the pace of the cost cuts. In March 2002, Clarke reported to Capellas that he had fallen short on efforts to reduce expenses. Walter Hewlett and other merger opponents seized on this and quickly made Clarke's comments the centerpiece of their legal effort to derail the merger. But when Clarke was dragged into court that April, he wouldn't give in; he testified that his clean-team managers were simply "sandbagging" the integration effort, trying to make themselves look good by giving him numbers they knew they could beat. He could wring the savings out of them, Clarke argued. If managers could hit his cost targets, he knew he'd set them too low. "It's a cat-and-mouse game," he says. "I would say, 'Good, now why not go and get me some more?'" Clarke's testimony helped Fiorina win the case--and his bargaining skills with clean-team managers helped HP deliver the cuts as promised.

The cuts, in fact, have helped HP turn two ailing PC businesses into a single, healthy one. At the time the merger was announced, HP and Compaq's PC businesses were losing a combined $372 million a quarter; the merged PC business is now profitable. While HP currently sells both Compaq and HP computers--allowing the company to grab more shelf space at retailers like Best Buy--the guts of the machines are identical, and manufacturing costs have been trimmed by 17 percent. (An accounting change also helped, though HP says its PC business would be profitable even without it.)

That's not to suggest there's any verdict as yet on the merger as a whole. Last November, Capellas resigned to become CEO of WorldCom, and in March, HP reported revenues of $17.9 billion, $600 million less than expected for its fiscal first quarter. Only HP's printing business grew last year; without it, HP would have reported a loss.

The real issue, of course, is how HP will perform once tech grows again. To ensure that it will, Fiorina has asked Clarke to tackle HP's next Mission Impossible: trimming another $1 billion out of the business of warehousing and shipping everything HP makes--printers, PCs, cameras, servers--and the myriad parts that go into these products. It's a challenge that Clarke relishes, and one that HP's suppliers have good reason to fear. "Jeff is like a bulldog on a bone," says Mike Winkler, who ran HP's supply chain before Clarke took over last December. "If he takes it, there's no way he's letting go." --BRIAN CAULFIELD