(FORTUNE Magazine) – In almost a quarter-century as a business journalist I have logged a lot of time in the company of CEOs, a few of them great--Sam Walton, Andy Grove, Jack Welch, Michael Eisner--and others, well, not so great. It's easy to tag most of the stars with some dominant characteristic that sets them apart. Walton was the world's greatest motivator. Grove's intellect towers over his peers'. Welch is the ultimate social architect. Eisner is a unique blend of business and entertainment genius. Such labels may not tell the whole story, but they make it easy to think and write about these leaders in interesting, instructive ways.

Roberto Goizueta of Coca-Cola, who died in October, was altogether different. He was the CEO I knew longest and best, and yet the one I still find most puzzling, the one most difficult to pigeonhole. I was saddened by Roberto's death, feeling a sense of both personal and professional loss. It's as if I'm some old sportswriter who has watched one of the greats--Montana, Aaron, Jordan--leave the game for the last time. I'm not certain I'll see the game played this way again, and I'll miss the pleasure.

If you caught any of the many eulogies published upon Roberto's death, you already know about his record as a wealth builder: He took Coke's market value from a mere $4.3 billion in 1981 to as high as $180 billion this year. So his place in the pantheon of truly great chief executives is forever secure.

Yet I remain puzzled. After years of interviewing the man, traveling with him, exchanging correspondence with him, arguing with him, and reading his essays, I still don't understand exactly how he managed to take a company written off by most people as a spent bullet in a mature industry and transform it into one of the glamour stocks of two decades. To try to get at an answer, I have excavated my old notes and tapes, and I have quizzed his close associates and confidants: Doug Ivester, his protege of many years and successor as Coke CEO; Don Keough, his chief operating officer at Coke for all but the past five years; and two powerful owner-directors who were with Roberto for most of the ride: Warren Buffett and Herbert Allen. What follows, then, is an attempt to unlock Roberto Goizueta's "secret formula" for being a superb chief executive. Warning: The quest uncovered a good bit of paradox and no small amount of contradiction.

THE DEVIL IS IN THE DETAILS, BUT--

For starters, there's the question of whether Goizueta was a hands-on detail man or an above-the-fray, clean-hands CEO. And the answer is clearly--both.

When he first came to power at Coke in 1981, many journalists and Wall Street analysts quickly concluded that Roberto was a boring engineer who was hopelessly bogged down in the details of his company and seemingly uninterested in all the problems on which they were focused, like, say, some super-duper ad campaign Pepsi had just launched. They were half-right, of course. He was immersed in the details of his company. He knew the per capita soft drink consumption rate in remote countries. He could quote the price-escalation clauses in high-fructose corn syrup contracts. And most important, he knew the cost of practically every cent of capital Coke had invested anywhere in the world and its expected rate of return.

What few of us realized was that while Roberto was marinated in all those details, he had no intention of attending to them himself. In fact, rarely has there been a CEO who delegated more of what we normally think of as the "work" involved in running a company. Nor did anyone appreciate just how brilliant his strategic plan for Coke was--even though he talked about it incessantly and took to passing it out in pamphlet form like an evangelist.

HAVE A REAL STRATEGY

I'll never forget my first meeting with the man, very early in his tenure. I was a young Atlanta bureau chief for the Wall Street Journal. He summoned me to his office, where I found him sitting behind his desk, suit jacket still on, smoking a cigarette. He introduced himself, took polite exception to my wearing a beard (he hated facial hair, I think, because of its associations with Fidel Castro, the man who robbed his family of its magnificent home and wealth), and said in his thick Cuban accent: "You and I need to get to know one another. You seem to be quite interested in the Coca-Cola Co., and things are going to be very different around here from now on."

Then he told me about his plan. Basically, he said, the company now had a strategy that would focus entirely on increasing long-term share-owner value (he hated the word "shareholder" because he said it wasn't precise enough). He planned to reduce the percentage of earnings paid out in dividends from almost 60% to around 35%, he explained, and for the first time Coke would take on debt. This would free the company to explore new opportunities and--very important--lower its cost of capital. He would divest businesses that weren't likely to pay off for shareholders, and there would be no sacred cows. Performance would be rewarded; perfect attendance would not. It was the only time a CEO ever explained to me a strategy so simple that it seemed almost naive, and I, along with everyone else outside the company, was skeptical.

At the time, remember, Coke was in disarray. It had lost significant momentum to hard-charging Pepsi and had yielded only a 1% compound annual return to shareholders for an entire decade. Its culture was one of entitlement and arrogance. And it was locked in a paralyzing war with its own bottlers. To compound these problems, the company had no communication with Wall Street--no analysts' meetings, no interviews, no comment--and unusually hostile relations with the business press.

Not surprisingly, the announcement that a shy, 49-year-old patrician Cuban virtually unknown to the outside world was assuming the helm of this famous but seriously screwed-up company had underwhelmed practically everyone. With the resume of a competent general administrator, he had no credentials as either a marketer or a finance man, yet he was planning to revive probably the greatest consumer brand in history? And here he was passing out these scholarly little pamphlets.

Of course, it all worked. Those who believed--a group that eventually came to include Buffett--and those who signed on as soldiers in Roberto's crusade became wealthier off soda pop than they ever dreamed they could, watching their stock multiply over and over and over for 16 years.

The simplest explanation of why it worked is that Goizueta was so diligent at performing the few primary duties he assigned to himself, the most important being the proper allocation of the company's capital. It's a job that doesn't require much heavy lifting--if you know everything there is to know about your company and you're very smart.

Roberto's calendar was remarkably uncluttered. Journalists and analysts were always amazed at how much time he took to respond, in detail, to the things they wrote or said. He was often available for a chat on the phone, a drop-by interview, or even lunch. His classically appointed office suite was eerily quiet and rarely showed any sign of activity. Usually when a guest arrived, "the Chairman," as he was known, was down the hall checking his stock ticker. And most afternoons he headed home around 4:30--riding up front with his driver and listening to country music--for a quiet evening with family.

PROMOTE THE BEST...

One reason Roberto could keep banker's hours was the decisiveness with which he made executive choices and the extraordinary trust he placed in those chosen. Shortly after assuming office, for example, he made two characteristic decisions. He quickly purged from the company one of the men who had been his rival for the top job, then anointed the other, Don Keough, as his right hand. Keough was everything Goizueta wasn't--gregarious, motivational, in-your-face, peripatetic. And for the next decade he would circle the globe in his Gulfstream, starting fires for Roberto and putting others out. He inspired the troops, fought Pepsi, launched invasions of new territories, bought and sold divisions, ran off underperforming executives, confronted recalcitrant bottlers, and schmoozed big customers. For these efforts Goizueta was happy to pay Keough more than many CEOs earn. When Keough retired a few years ago, he took with him stock now worth around $400 million.

...AND SHOWER THEM WITH STOCK

Herein lies an essential ingredient of Roberto's formula. Once he felt he had the best possible people in executive jobs, he rewarded them lavishly with a stake in the game: Coke stock. "The compensation system was the key to driving shareholder value," says Herbert Allen, chairman of the Coke board's compensation committee. "It completely aligned bottom-line productivity with the stock price."

And while Roberto was happy to delegate many typical CEO responsibilities to others, he reserved for himself one task often delegated to administrators--the annual award of stock options. Thus when Coke's 35 or so top executives would gather periodically for a dinner at, say, Atlanta's Capital City Club, everyone in the room was a millionaire several times over, and they all knew from whom their wealth had come.

THINKING IS WORTHWHILE

Okay, we've identified a few of the duties with which Roberto occupied his time. So what else was he doing? "Thinking" seems to be a big part of the answer. As Keough puts it, "He believed a big part of his role was to make sure he had a viable long-term strategy for the company, and he wanted to have time to think."

What kind of thinking? According to Doug Ivester, it was consistently linear, based on his Jesuit training as a boy in Cuba. "He always began with the broadest possible statement everyone could agree on," says his successor, "and you never moved on to step two until you had exhausted the possibilities of step one."

As Warren Buffett puts it, "Roberto was super-strategic. He wasn't ever focused on anything other than getting results at Coca-Cola over a long period of time." Just a little over a year ago, I stood with Roberto on a stage in New Orleans in front of a roomful of CEOs, gathered for the FORTUNE 500 CEO conference, and heard him exclaim for the 10,000th time, "A publicly traded company exists for one purpose and one purpose only: to increase share-owner value. If it does that, all the other good things will follow."

Roberto Goizueta was way ahead of his time in adopting this point of view. He was also a pioneer in figuring out how to do it. Early on in his CEO tenure, he spotted Ivester, a former outside auditor for Ernst & Whinney, and enlisted him as his collaborator in the financial reengineering of the company to enhance share-owner value. Before anyone had ever heard of EVA, they were examining the underlying cost of the company's capital and redeploying it. When he sold Coke's wine business in the early '80s, Goizueta explained the move in terms even I could understand: "If we earn 3 cents for every dollar of capital we have invested, this is fundamentally a bad business to be in," he said. "We would be better off to take the $250 million we have invested in the wine business and put it in the bank." When he bought Columbia Pictures early on, it wasn't one of those hang-with-the-starlets decisions. He almost bought a pharmaceuticals company but found the earnings of the movie business more appealing for his needs at the time. And even though Coke proved it knew nothing about running a movie studio, it proved great at selling one--to poor, star-struck Sony.

While Keough was out strong-arming Coke's bottlers into accepting new rules of the game or selling their franchises to Coke, Goizueta and Ivester were back home figuring out ways to boost Coke's stock by spinning off the capital-intensive bottling assets into a new company. And long before stock buybacks became all the rage, they decided Coke stock was a great investment for the company. Ultimately, they bought around a third of the company back for an average of about $11 a share (Coke was trading at around $65 when FORTUNE went to press).

There was one other important ingredient to Roberto Goizueta's formula, a hard one to acquire if you don't already have it: a taste for risk. As Keough puts it, "He had a hell of a lot of guts."

ROLL THE DICE

Think about it. He was the first person ever to stick the Coke brand on a derivative product, bucking years of company orthodoxy that said it would destroy the franchise; Diet Coke went on to become the most successful consumer product launch of the '80s. He changed the sacred formula of Coca-Cola, a move that would have been previously unthinkable; New Coke blew up in his face and became the biggest consumer product disaster of the '80s, but he quickly reversed his decision and, in fact, capitalized on all the attention it got.

I don't know. Maybe none of these ingredients really hold the secret of Roberto Goizueta's extraordinary success. Maybe it's all alchemy. But I do know that what the man did is worth studying. He was born rich, lost everything, did something he loved at one company for his whole life, and made a billion dollars for himself and more than $175 billion for others.

He could be funny. "You hope you don't hire anybody who is stupid," he said at Herb Allen's Sun Valley conference last summer, "but if you do, pray that they don't have a lot of energy." He could be ruthless, once telling FORTUNE: "A public flogging is necessary every now and then. It focuses the attention of the troops."

That, says Keough, was perhaps Roberto's greatest skill: "the ability to focus the entire Coca-Cola system, including me, for 24 hours a day, on building shareholder value." Buffett puts it another way: "Coca-Cola was Roberto's painting. It was never finished, and he was never totally satisfied with it. But he had that Sistine Chapel ceiling in his head, and he was always working on it."