(FORTUNE Magazine) – Larry Bossidy, CEO of AlliedSignal, has been thinking about e-commerce. It may be accurate to say that Bossidy has been obsessing about it. "Every company," he says, "has to have an Internet strategy. You've at least got to look at the Internet for distribution. If you don't, somebody's going to come along and do a Dell Computer on you and destroy your business."

A week before saying this, Bossidy put on a miniconference for the leaders of his company's business units, which create aerospace and automotive products, chemicals, fibers, plastics, and advanced materials. He invited two outside speakers: Michael Dell and John Chambers, the CEO of Cisco Systems. At day's end Bossidy gave each business unit head two months to report back with an e-commerce strategy for his business. Those plans are due a couple of weeks after this column runs.

The Internet will change everything; everyone knows that now. A year ago I was talking to an executive of a high-tech company founded in a Palo Alto garage that is now a landmark. "Don't tell anybody," he said quietly, "but Dell is forcing us to reorganize our whole sales end." Today that's no secret: His boss, Hewlett-Packard CEO Lew Platt, talks about it in Michael Dell's presence. And Platt has announced he will split up his company, partly to get HP's computer operations up to industry speed as defined by Dell. Less clear than the fact of change is what exactly it is that the Net will change, and how and how much. Downright murky is what's most important in the long run: the management processes businesses must have to thrive as e-corporations.

It's easiest to see the direction of change by starting where Bossidy does--with distribution, the physical movement of physical stuff: loading docks, warehouses and trucks, products on conveyor belts, pallets in warehouses. True, business growth is strongest in knowledge products and services, which are intangible. But manufacturing and distribution processes have remained the proving ground for new ideas, probably because they are clear, linear, and countable, and because most of the time (though not always) their lessons apply to the intangible economy.

Dell's business model is as easy to understand as it is revolutionary. Traditional distributors do three value-adding things. They aggregate information (names of buyers and sellers, product specifications, knowledge of supply and demand); they shoulder some quality and credit risk for both buyer and seller; and they aggregate inventory, assuring prompt delivery. (The words a shopper least wants to hear: "We have to order that from the factory.") Dell does all three value-adders but vaporizes the distributor. It uses the Web to aggregate information, instantaneous electronic payment to mitigate credit risk, and build-to-order manufacturing to assure availability. Typically a Dell customer's computer is assembled to his specifications, loaded with the software he requests, and ready to ship within two hours of receipt of the order--faster than some companies can get finished goods out of a warehouse.

Says Bossidy: "If this works for computers, it's going to work for automobiles, furniture, carpets, appliances, anything." Any dealer is dead if his only claim to a share of the money in the value chain is a window in front and a warehouse out back. The Web gives me a window in your den.

So we know that there will be fewer distributors. Many moms-and-pops will die. This is a big deal. Anybody who has ever tried to crack Japan knows that distribution is everywhere and means damn near everything, same as the body's distribution systems, nerves, and blood vessels. As for the warehouse: Convert it to condos. The amount of inventory in the economy has worked its way erratically but inexorably lower, like Edward Bear going downstairs--bump, bump, bump. The inventory-to-sales ratio is now less than 1.4 to one. A decade ago it was above 1.5 to one, according to Commerce Department data. As inventory shrinks, money tied up in materials, storage, and financing slips its ropes. Up for grabs in the supply chain, says James Duffy, CEO of Benchmarking Partners, a Cambridge, Mass., consulting firm: $1 trillion a year, give or take a few billion.

Want some? Says Bossidy, "We do this exercise: 'Who makes money on our product? Do they deserve it? Why aren't we making that money?' You find remarkable stories of how much other people make off your products. And if you can't come up with a positive reason they deserve to, you go after it."

The Net won't kill off intermediaries, as some say. Unintermediated Internet is as unwieldy and inefficient as unintermediated inventory. But e-commerce will devalue a lot of what middlemen were paid for, wrecking pricing for plain-vanilla distribution and those who depend on it. "It's a bit obscure what they're going to do," says Bossidy--that is, how dealer-distributors can differentiate, and thereby earn a living wage. They might try using scale and service (logistics, installation, and repair, for example), or might convert themselves from transshippers to assemblers, as Ingram-Micro has in personal computers--thereby, incidentally, becoming a rival of its suppliers. It's also unclear how suppliers will handle the situation: Do they cut themselves loose from dinosaur distributors (and alienate their sales channel) or help them evolve (and run the risk that they will get clobbered by some quick mover who has less to lose)?

This battle is part of the ongoing war: the one about finding and keeping profitable customers. E-commerce is not just a new front in that war--it's a different kind of front, where every unit can be both on the front lines and in the rear. Phone the New York Times and ask for home delivery: You are given the name of a local distributor. Log on, though, and you can find the Times distributed by America Online or offered direct at www.nytimes.com. The consumer chooses, because he holds the mouse.

Offline, people talk about "upstream" and "downstream." Online, the word is "clickstream," the sequence of mouse clicks as a Web user navigates from one page to another. The Web is an infinitely branching fractal, like a river delta. If you're too far down the clickstream, any one of a zillion other opportunities may distract me before I ever get to you. You can pay to have people channeled toward you, the way the Army Corps of Engineers channels rivers, but that gives back your piece of that trillion bucks and runs the risk of cutting you off from the ecological richness of the delta. It also puts you back in the distribution game that e-commerce can get you out of, and with an added problem: The Web is so fluid that control over channels is likely to be fleeting--or very expensive.

So how do you move up the clickstream? The answer--rather, the way you learn the answers--is to improve your ability to learn with and about customers. There are a number of "knowledge processes" in business. "Customer learning" is one that gets little management focus--but I'd argue that it's the sine qua non of success in electronic commerce. Why? First, because the e-commerce battle is fought with information, not with inventory. Michael Dell explains his success simply: "We substitute information for inventory." The closer you are to knowing about buying decisions in real time, the better your chances of getting dibs on part of that supply-chain jackpot. "Daily customer information--that's a gold mine," Bossidy says. Today in the auto industry there's a multifront war between Web-based car-buying services like Autobytel.com, dealers who put up their own Websites, and automaker-sponsored online-dealer networks. Their weapons are chiefly information: The manufacturers' sites won't let others know about incentives, for example; the car-buying services counter with a package that includes information about all makers' marques.

The high ground on this battlefield is not so much information as it is mutual learning: creating a process whereby the seller and customer learn with, from, and about each other, and whereby customers can learn from other customers. A well-developed, two-way customer-learning process will make me yours and keep you mine. We'll explore that next time.