(FORTUNE Magazine) – A cold rain made it easy for the boys of Troop 184 to pretend they were soldiers. In fact they were twoscore American Boy Scouts--sons of ex-pats--on the beaches of Normandy, where, perhaps, a grandfather or great-grandfather had fought 55 years ago. Warm and dry in London a few days later, their scoutmaster, Kent Greenes, said: "We did this five years ago. We want to do it every five years, so we remember. So we have this history burned in our brains."

I was talking to Greenes because of something said a few weeks before by Ron Shevlin, a senior analyst at Forrester Research, at one of the many conferences on knowledge management. Most of Shevlin's audience seemed to be people who sell, buy, or operate groupware, data-warehousing and intranet software, human-resources systems, etc.--the retorts and alembics that companies use to collect and share know-how. Knowledge management is the hottest thing since reengineering, yet a Forrester study had shown that six out of seven companies investing in it are doing so on faith--not even trying to measure the return. That won't last, Shevlin warned: "There's six to 12 more months of easy money" before KM gets scrutinized like anything else. The gravy train has sprung a leak, as it should. There's no more ardent advocate of knowledge management than I, but show me the money or kill it.

Greenes is, as best I can figure, knowledge management's top moneymaker. In 1997, John Browne, CEO of British Petroleum (now BP Amoco), asked him to lead an effort to improve performance by sharing best practices, reusing knowledge, accelerating learning, and so on. Greenes and nine teammates used that first year to bone up and take a few practice swings, netting $20 million. In 1998 they went to work in earnest, and in one year brought $260 million to the bottom line--documented savings attributable to knowledge management--plus $400 million more likely but not yet booked. Nearly $700 million. How'd he do it? The way he led his Scouts: by making war stories come alive.

BP Amoco isn't big on theory; it's a company of engineers, roustabouts, and finance guys, their ethos shaped by places like Alaska's North Slope, where Greenes himself ran the geoscience development of an oil field in 1989-90. So his first act in 1997 was to find results--a story about somebody who had done knowledge management without knowing it. He found his tale in the Schiehallen oil field, a North Sea field considered too expensive to develop until a team spent six months pestering colleagues to share cost-saving tips. They were called wimps for not rushing out to "make hole"--but the learn-before-doing approach saved so much time on the platform (at $100,000 to $200,000 a day, not counting drilling costs) that they brought the field into production for $80 million less than anyone thought possible.

Thus armed, Greenes began knocking on doors, looking for customers. At flat, decentralized BP Amoco, the 126 business unit heads sign "performance contracts" for financial, environmental, and other results and then are left pretty much on their own. So Greenes looked for people who had tough targets, needed help, and weren't afraid to ask for it.

Two key principles of knowledge management are revealed here. First, Browne told Greenes, "engineer your pilot projects for success" by looking for people who want to make them work, avoiding cynics or suck-ups. Greenes' rule: He paid only for his own team's expenses. Everything else--money and top people--had to come from the business unit. Second, every company has a salient cultural trait you can grab onto. At BP Amoco, performance contracts have the gravity of an oath; people will do anything rather than fail to deliver. "I leveraged the hell out of that," Greenes says.

For 1998, Greenes assembled a portfolio of 15 projects where knowledge management could make money. Among them: helping the company enter the Japanese retail market; reducing downtime at a polyethylene plant; doing a "refinery turnaround"--a scheduled shutdown and refurbishment--in Rotterdam.

Lots can happen in the four or so years between refinery turnarounds. People leave, taking experience with them. New technologies come online. At another refinery, two years ago and ten time zones away, someone invents a faster way to do a task. So Nerefco, the Dutch refinery, got a "peer assist"--a facilitated two-day meeting when people who had done turnarounds briefed its team before the project began. The tips they got saved $9 million. These discussions need a leader in order to stay focused. Explains Greenes, enunciating another key principle: "If you say, 'Tell me everything you know about X,' people resist, not because they don't want to share but because it seems like too big a job. But if you ask in a specific enough way, they're happy to tell you."

The "peer assist" is one of five tools--"real simple stuff," Greenes says--that his team used. The "after-action review" (adapted from a U.S. Army practice) is for learning while doing, by taking just 15 minutes to ask four stock-taking questions: What was supposed to happen? What actually happened? Why is there a difference? What can we learn and do from this? "The retrospect" is a more elaborate postmortem. Useful in and of themselves, these three tools also are used to create a Web-based folder of artifacts, videoclips, checklists, and most important, e-mail hyperlinks to people who have stories to tell. Greenes made a score of these knowledge assets, with titles such as "What does BP Amoco know about restructuring?" "...about building retail sites in new markets?" "...about refinery turnarounds?" and so on. "They are not bullet points," Greenes emphasizes. "The most important part of any knowledge asset is the link to the people who were involved."

Connection, not collection: That's what knowledge management must offer if it's to make money. Greenes' other two tools use technology to connect people. One is actually named "Connect": a voluntary intranet Yellow Pages to make it easier to find expert help; more than 12,000 employees (out of 100,000) have put themselves into Connect. The other is a snazzy piece of technology that delivers desktop videoconferencing, multimedia e-mail, a real-time shared whiteboard, and other goodies that let someone in Venezuela sit at the virtual elbow of a colleague in Alaska. "Virtual Teamwork" is the only expensive part of Greenes' operation, "but what made it work," he says, "is that we spent as much money on coaching and training as we did on the technology."

Greenes invented almost nothing: The BP Amoco tool kit is a living testament to the megamillion-dollar value of reusing ideas. Greenes figures that between 30 and 40 of the BP Amoco business units have internalized the knowledge-management tools and ideas, and more are coming along fast. More than 300 people have become part-time volunteer knowledge managers. Says Greenes: "I think that I've worked myself out of a job."

So next month he hangs up his hardhat and heads to San Diego, to reuse his experience as the chief knowledge officer for the commercial and international operations of SAIC, the employee-owned high-tech research and engineering company (1998 revenues: about $4.6 billion). A few things--such as Connect--will continue to be run centrally. For the rest, Greenes hopes that 700 million bucks is the kind of history that burns in the brain.