(FORTUNE Magazine) – Try dropping this little suggestion in your next strategy session: It's time we got busy eating ourselves.

The word "cannibalism" is on people's lips these days, and it has nothing to do with the much publicized return of Hannibal Lecter. Rather, it describes a predicament--one that is rapidly emerging as the managerial dilemma of the e-economy. Breakneck change and a raft of Internet upstarts are threatening to overturn long-successful technologies and business models. There are two choices: Yield to your instincts and protect those still-profitable technologies and models. Or preemptively overturn them yourself, even if it means eroding the very revenue streams upon which your company is founded.

It's a devil of a dilemma, but the upshot for corporate America is fairly simple: Companies that learn to cannibalize themselves today will rule tomorrow's business jungle. Those that don't will find themselves in someone else's pot.

If eating one's own sounds like a hard thing to do, it is. It's counter-instinctual. It means embracing technologies that will destroy the value of past investments--whether they be factories, relationships with a distribution channel, or psychic investments in how things ought to be done. It often means purposefully depressing profitability and share price. It means, in short, doing things that every good manager is trained not to do.

Indeed, when two business school professors recently surveyed 11 marketing textbooks, they found that nine of them portrayed cannibalism as something to be avoided--"an unfavorable consequence of...new product innovation," in the words of one. Another article warned against "the error of cannibalism." This taboo made perfect sense, of course, when cannibalism was a matter of Miller Lite stealing market share from Miller High Life, or a new Bob's Big Boy stealing customers from the Bob's Big Boy three miles down the road.

But when it comes to radical, disruptive innovation--innovation "which strikes not at the margins of the profits and the outputs of the existing firms but at their foundations and their very lives," in the words of the Austrian economist Joseph Schumpeter--not cannibalizing oneself can mean surrendering any hope of surviving the next wave in anything but greatly diminished form or at best as the vassal of an acquirer.

The Internet is that kind of innovation, but it's not the only one. For Eastman Kodak, embracing digital imaging means undermining a century's worth of specialized investments in the production, processing, and distribution of silver-halide film. For Detroit, embracing electric vehicles could spell the same for the internal combustion engine. Even so, it's the Internet that's making the cannibalism imperative widespread. Jack Welch refers to General Electric's Internet business units as "destroy-your-business.com." Intel's Andy Grove talks about a "valley of death" that incumbents must cross. Harvard Business School's Clayton Christensen calls it "survival by suicide." Pick your metaphor. The smartest managers are already doing it.

"Someone once asked me what causes me to wake up in the middle of the night," says John McCoy, CEO of Bank One. "It used to be bad loans. I don't worry about bad loans anymore." He worries about the Internet--and specifically the wave of Net startups offering online banking, online mortgages, online credit cards, you name it. As McCoy sees it, merely harnessing the Internet to enhance Bank One's existing business isn't protection enough against the invaders (though bankone.com does a brisk business). Instead, he launched WingspanBank.com (tag line: "If your bank could start over, this is what it would be"), a freestanding all-Internet bank that will be free to poach Bank One's customers with the lure of higher deposit rates and other goodies that only an unencumbered "pure play" can offer. Explaining the decision to Wall Street, McCoy said he was "willing to cannibalize existing business to build new business"--even if it meant eventually having to shutter branches.

Likewise PetSmart. Following the lead of Barnes & Noble and Toys "R" Us, the pet-supply chain recently spun off its online venture as a separate company, partnering with the Internet incubator Idealab to launch PetSmart.com, in which it retained only a 49.9% stake. "They are fully willing to cannibalize their existing business," enthuses Idealab Chairman Bill Gross, "including letting the online venture recommend people not to go to the stores, including letting the online venture price lower than the stores," even if that creates a gray market in its goods.

Even Sears, hardly a trailblazer on the Web, is making munching noises. "If you accept this transformation is going to happen--and I submit it's going to happen--you have to get over this issue of cannibalization," says CEO Arthur Martinez. "We should be indifferent to what channel the customer picks." So what percentage of Sears customers will end up picking the Net over Sears' physical stores? "TBD [to be determined]," Martinez says. "But if you wait for the answer, it's Game Over."

These are managers who have decided to allow the gale of "creative destruction"

--Schumpeter's famous force, which "incessantly revolutionizes the economic structure from within, incessantly destroying the old one, incessantly creating a new one"--to blow freely within the walls of the company. They recognize that doing so may be the only way to keep customers within the firm, albeit at a lower return. They recognize that if they're to have any chance against the attackers, they must play the role of both defender and attacker.

They recognize that these days you can't have your company if you don't eat it too.

With his 6-foot 1-inch defensive lineman's frame, David Pottruck doesn't look like a cannibal, but he is one. In 1996 the co-CEO of discount broker Charles Schwab established a separate online unit, e.Schwab, with its own staff, own offices, and own sense of mission. Then he did the unthinkable: He let e.Schwab eat Schwab.

The moment of truth came in late 1997, just as demand for e.Schwab's $29.95 online trades was booming beyond anyone's expectations. Problem was, customers with Charles Schwab's traditional brokerage still had to pay an average of $65 per trade. The two-tiered pricing system was clearly awkward: Some customers were keeping small sums of money with Charles Schwab to maintain access to live brokers, then executing their trades through e.Schwab. So Pottruck came to a radical decision: All trades would be priced at $29.95. In essence, all of Schwab would become e.Schwab.

Employees in the company's branch offices were skittish. "All of them thought they would have no more business and were going to lose their jobs," Pottruck says. "It attacked our old business." Schwab's board had its doubts too. The price cut would shave an estimated $125 million off revenues, and the company's stock would clearly take a pummeling. Even Pottruck himself wasn't quite sure of what he was doing. "I can't tell you honestly that I didn't lose a lot of sleep about it," he says now. "I thought, 'Maybe all we're going to do is give away revenue, profit, and share price.' " He was reminded of the time he and his three brothers went on a rafting trip and ended up on the far bank of California's Clavey River just above a waterfall, needing to swim across. "The question was," he recalls, "could we reach the other side of the river before being swept down the waterfall?" The four of them stood there, trying to gauge distance and current, until one of his brothers finally plunged in.

To prepare staffers for this particular plunge, Pottruck staged an odd little ceremony in which he walked across the Golden Gate Bridge with 100 or so of Schwab's top managers in tow. He called the event Crossing the Chasm--symbolism intended to suggest leaving one business model behind and embracing a new one based on the Internet. Then he took his entreaty to the board: "It's not going to get any easier, so we might as well move forward and take the pain."

In January 1998 the price cut took effect. Schwab's stock lost almost a third of its value. But the short-term pain yielded outsized long-term gain: Total accounts climbed from three million to 6.2 million; the stock recovered; $51 billion in new assets poured in during the first six months of this year alone. (Pottruck's brother made it safely to the far side too.) More incalculable is the value of grabbing 42% of a new market that would otherwise have been dominated by upstarts like E*Trade. "You can't steal second," says Pottruck, "with one foot on first."

But why is Pottruck such a rarity? Why aren't there more examples of successful cannibals?

That's the question that consumes Clayton Christensen, a preternaturally earnest, 6-foot-8 Mormon who left his life in the high-tech business at age 38 to earn his doctorate in management. On his office shelf at the Harvard Business School, where he's now a tenured professor, are some artifacts of his dissertation research: a row of six disk drives, each one roughly two-thirds as large as the one next to it--a reminder of the ineluctable march of innovation and, more important, the upheaval it causes. Each major innovation in the disk-drive industry, Christensen's research showed, caused the current dominant player to fumble its leadership.

When Christensen published these findings in his 1997 book, The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail, few took notice. But it's as if he'd chosen the fall of 1913 to publish a survival guide to trench warfare; when the Internet began roiling industry after industry with exactly the sort of dynamic he documented, the book vaulted into bestsellerdom. Now the speaking and consulting requests roll in at the rate of 50 or more a week, and Christensen wears the hangdog look of a man who could stand reminding that today is Monday and he's in Boston.

What makes Christensen's sudden celebrity on the corporate circuit so improbable (he's an adviser to Intel's Andy Grove, among others) is that his message to managers is so...well, the word "apocalyptic" comes to mind. As in: Old-line companies, when faced with "disruptive" technologies that make existing ones obsolete, are pretty much doomed to lose the market to new entrants. None of the leading makers of vacuum tubes, Christensen bleakly notes, became players in the transistor market. Not one. Ditto for integrated steelmakers confronted with mini-mill technology. Ditto for big department stores faced with the rise of discounters. Ditto for...okay, you get the point. All those innovations caused miniature extinctions; the Internet could cause a very big one. Calling in Clay Christensen to talk to your company can feel a bit like hiring the Four Horsemen as turnaround consultants.

So why do big-company managers seem to suffer from this "incumbent's curse"? Because they're hidebound, obtuse, fat, and complacent, right? Wrong. Christensen's thesis: Corporate managers end up shunning important new technologies and markets precisely because they are being "good managers." That is, they are listening to shareholders and customers, "focusing investments and technology on the most profitable products that are currently in high demand by the best customers," he writes. In a way, they suffer from an excess of rationality.

These rational instincts serve managers well when it's a matter of incrementally improving existing offerings. Yet when it comes to disruptive innovations like the Net, Christensen writes, "there are times at which it is right not to listen to customers, right to invest in developing lower-performance products that promise lower margins, and right to aggressively pursue small, rather than substantial, markets" (see box). The more successful the company, the harder this becomes. Like many companies in this hot economy, Schwab was still growing quickly and had a few billion good dollars left in its old way of doing business.

Balancing two colliding imperatives--getting the new business model up and running while keeping the old one from crumbling under you--might be too much to ask of one managerial mind, Christensen concludes. Which brings him to the prescription for which he's become renowned: The only way for companies to survive such transitions, he says, is to set up a separate company to cannibalize the parent.

He's quite serious about this. "Every day," explains Christensen, "everybody in the company has to make decisions about what they're going to put first and what they're going to put second. For top managers, it's 'Which projects do I fund?' For the sales force, it's 'What products do I push and what products do I not push?' For engineers assigned to two development projects at once, it's 'What am I going to work on today?' " Without a separate entity that has its own cost structure and set of customers, he says, the answers to such questions will reliably kill off new projects that "don't belong." Strategy determines structure, as the saying goes, and two strategies call for two structures.

Once separate, should those two structures be permitted to cooperate? "It's commonsensical," replies barnesandnoble.com spokesman Ben Boyd, slightly provoked at the question. "We have 520 stores. Why in the hell would we not try to leverage that to the betterment of the online operation?" A reasonable point. But what's most interesting is his choice of pronouns: we. Cooperation is okay, but only insofar as it's done through arm's-length, Adam Smithian transactions, with each party pursuing its own, and only its own, self-interest. As soon as executives try to yoke the strategy of the spin-out to the strategy of the mother ship--as soon as "us and them" becomes "we"--failure is all but ensured, Christensen cautions.

Which makes of particular interest the recent efforts of some corporations to do just that.

It's a bit strange watching John "Launny" Steffens, vice chairman of Merrill Lynch and chief of its 14,800 brokers, standing in Merrill's walnut-tabled boardroom, peering into a laptop, trading stocks over the Internet. This is the same Launny Steffens whose most famous utterance, spoken just over a year ago, was that online trading "should be regarded as a serious threat to Americans' financial lives."

Today, however, Steffens is Mr. Online Guy. "We're going to sell 100 shares of AOL," says Steffens as he points, clicks, and sends his order into the ether. (That his own firm rates AOL a "buy" is beside the point here.) A very pregnant pause follows, then a dialogue box pops up: "Your Request Is Processing. Please Wait."

Waiting too long, of course, is precisely what Merrill stands accused of, though the delay isn't all that surprising, given that its bout with cannibalism promises to be much more painful than Schwab's. While Schwab was already a discount brokerage that was not so far removed from being an entrepreneurial company itself, Merrill is a full-service brokerage entering an arena that has come to prominence largely by bashing full-service brokers. (Then, too, Schwab's brokers are salaried, while Merrill's work on commission.) "If we didn't make these changes," Steffens says of the turnabout, "we were going to have more of our business at risk than we really thought or understood."

But while Steffens may have eaten his words, is Merrill ready to eat itself? Some clues might lie in the events leading up to the June 1 announcement of its Net strategy. Herbert Allison, the company's wonkish president and heir apparent (emphasis here on apparent), had reportedly been pushing to set up a separate online unit that could compete with the core brokerage, a la Clay Christensen. But Steffens, himself a former broker, had insisted that any Internet initiative fall squarely under his watch. Steffens prevailed, and by mid-July, Allison was out of a job. (Reached at his home in suburban New York, Allison declined to comment.)

It's easy to see why Steffens wanted control. As it is, his brokers are only barely quiescent. "We're going to be forced to attract more clients to [earn] the same amount," says one of a number of brokers who spoke to FORTUNE with a request of anonymity. Weaker brokers who don't reach a certain asset threshold quickly, he says, will likely fall by the wayside.

Steffens' solution is, not surprisingly, a precarious sort of hybrid. Customers can choose from a confusing array of five different options, including unlimited Internet trading for a percentage of their assets and, beginning Dec. 1, a la carte $29.95 trades. "Many incumbents focus on hybrids," says Rajesh Chandy, a marketing professor at the University of Houston who studies cannibalism. "It puts them on the learning ramp of the new technology while giving them some time to leverage existing technology." But this hybrid is clearly fraught with conflict. "They're going to fight every day to balance the two different sides of the house," says Bill Burnham, who follows the online brokerage industry for CS First Boston. "It's like asking someone to go to work every day and put their friends out of a job." Already Steffens says that brokers won't be expected to promote the $29.95 offering, conceding that "they are not going to be a completely objective adviser" to clients.

"The question is," says Burnham, "will they ever aggressively promote the online business for its own sake? Or will it perennially be the stepchild to the advice business?" He answers his own question: "Merrill will never be able to fully take the gloves off and go mano a mano with the other online brokerage firms. Every time they want to grow the online business, they're going to have to consider how it's going to affect the existing business."

How do we know for sure that Steffens et al. are worried about such issues? They've called in Clay Christensen.

Cannibalism gets even more complicated when you don't own the distribution channel you're about to cannibalize--as Compaq Computer can attest.

Pity Compaq. Wall Street is convinced that rival Dell Computer's model of selling PCs directly over the Internet is sublimely correct. Compaq, meanwhile, depends on an old-fashioned network of resellers for roughly 75% of its sales. Let's see what happens when Compaq tries to become more like Dell.

Last November, Compaq announced it was aggressively expanding its own direct-sales program, DirectPlus. Predictably, resellers flipped out. "When they first showed it to us, we just hated it," says Steven Harper, president of a small reseller in Hutchinson, Kan., and chairman of Compaq's Small and Medium Business Advisory Council. Suddenly Harper's customers could bypass him and buy computers directly over the Web for 8% to 9% less than what he charged. Compaq's move "basically set our price," Harper says. "It forces us to play this little cat-and-mouse game: 'I wonder if my customer knows the price on the Web.' "

Not only would Compaq's new Prosignia line be available online, but the company's ad campaign seemed to give short shrift to the fact that buying from a reseller was still an option. Ted Warner, president of Connecting Point, a $12 million reseller in Greeley, Colo., was angry enough about it that he started steering customers toward Hewlett-Packard hardware instead. In the following months, Compaq's sales in the crucial small and midsize business segment swooned. Compaq blamed the dip on industry problems. But to Kirsten Campbell, a PC industry analyst at Merrill Lynch, the link was obvious: Compaq was losing the hearts and minds of its resellers.

In a series of heated meetings with Compaq vice president Michael Pocock, the resellers let their displeasure be known. They felt Compaq was obsessed with Dell and amnesiac toward the resellers that made Compaq the company it is. Compaq's dilemma: Should it plow ahead with the new channel and risk a collapse of sales in the old? Or should it backpedal?

"I think you've just got to bite the bullet and take six months of low sales," says Lester Thurow, professor of management and economics at MIT. "It is the valley of death. For at least one year, you report a disaster.... But it's obviously clear that Dell's doing it right. I don't think there's any other choice."

Christensen's reaction: "It's easy for academics and journalists to trivialize how difficult this is," he says. Nonetheless, "the evidence is just overwhelming that companies that try to manage the health of their distribution channel won't end up on top. If the channel is going to be disrupted, that's a force more powerful than any company." In other words, damn the resellers and full speed ahead.

Compaq's choice? Backpedal furiously. It temporarily suspended the right of some third-party Websites to sell Compaq products, and introduced new programs that would help resellers set up their own online storefronts and pocket an "agent fee" for steering business through DirectPlus. Then-CEO Eckhard Pfeiffer called the approach "customer choice," hoping to reaffirm Compaq's role as "the No. 1 channel-friendly vendor in the industry."

Places like Compaq, of course, argue that such hybrid approaches--what Pottruck calls "clicks and mortar"--are not a stopgap at all but an end state. They point to the splendid synergies that can be wrung from an artful combination of old and new: Unlike eToys, for instance, Toys "R" Us can use its stores as return centers for toys purchased online. "We have a huge competitive advantage compared to the pure-play guys," says Sears' Martinez. "We think there's a virtuous cycle between the online world and the physical world." Spend any time at Merrill Lynch or Barnes & Noble, and you'll hear the same argument in 1,000 variations.

You'll also hear a lot about how Internet sales aren't cannibalistic at all but additive--"plus business," as the saying goes. Estee Lauder reports that more than 30% of orders on its Bobbi Brown Website come from people who say they've never used Bobbi Brown cosmetics before and that more than 70% of online orders come from outside a 20-mile radius of a Bobbi Brown distribution store. That may be. But there's also some wishful thinking going on here. Research firm Jupiter Communications estimates that only 6% of Web sales in 1999 will be plus-business, meaning the other 94% will cannibalize traditional channels. (Merrill's experience bears this out; it reports that 84% of its new online assets are "conversions" from old accounts.) Those statistics, advises Jupiter, "should terrify traditional merchants."

Terror, in the end, may not be enough to overcome resistance to cannibalism. Korn/Ferry International, the blue-chip headhunting firm, knows that the onslaught of Internet entrants in its industry is exerting radical pricing pressure, putting customers ever more in control of the talent hunt. A consulting firm informed Korn/Ferry that fully half of its business could disappear within a few years. That's terrifying. But the company's Internet service, FutureStep, hasn't taken off, perhaps because it's charging clients exactly what they'd pay a full-service recruiter: one-third of a hire's first-year compensation. Why won't Korn/Ferry cut the price? "The problem is," HotJobs.com CEO Richard Johnson opined to Online Recruiting Strategist, "Korn/Ferry's recruiters don't want FutureStep priced below their services, because they believe they'll be cannibalizing one from the other. The company needs a different sales force to sell this."

Meanwhile, TMP WorldWide, a once-obscure recruitment advertising agency that also owns the fast-growing job board Monster.com, is pouring resources into a sector that promises to gobble up its traditional business. Impressed, Salomon Smith Barney analyst Lanny Baker has praised TMP as a "shareholder-friendly cannibal," noting that "few traditional offline companies in any industry" have been able to pull off such a feat. Will the TMPs of the world overtake the Korn/Ferrys simply because they're more willing to cannibalize?

Look at it this way. Industry after industry has thrived by not letting customers have control. "[We've] had this charmed existence for 30 years," Korn/Ferry CEO Windle Priem says of his industry. "We could take as long as we wanted, charge as much as we wanted." The Internet will end that, like it or not. Customers will have control--and companies are better off voluntarily surrendering the keys to the city now, before customers have fled to born-on-the-Web competitors. As Chandy puts it: "Cannibalize before there is nothing of value left to cannibalize."

This is brave stuff, and no one's saying it's going to be fun or easy. "These are the best people in our economy who are wrestling with this problem," Christensen says softly. "We should pray for them."