By Ram Charan
FORTUNE — It’s becoming an epidemic: the dead-on-arrival CEO who is doomed from day one because he or she was the wrong choice. Look at Yahoo (YHOO), which just got its fifth CEO in five years, or think of Leo Apotheker, who lasted only 11 months at the top of Hewlett-Packard (HPQ).
Investors blame the CEO when he or she flames out, but the real culprit is the board. The directors blew their most important job: making sure the company always has the right CEO.
To avoid such damaging failures, directors must seize control of CEO selection and pursue the task in a way that’s fundamentally new at most companies. I’ve analyzed 82 CEO failures from the past 20 years and have been on the scene of many successions, good and bad. I’ve observed what works and what doesn’t. The winning approach is clear, and more boards should go firmly on offense and follow it. The following steps are critical:
Reverse the usual process.
Most boards doom their efforts at the start by committing the blunder of considering CEO candidates without knowing what they’re looking for – especially the specific skills and relationship the company will need most. Directors need to think in the opposite direction. They should first define rigorously and specifically the unique requirements for the company and a successful CEO in today’s environment. What are the non-negotiable criteria the candidate must meet?
MORE: Business’s real problem: Uncertainty
That’s what IBM’s (IBM) board did in 1993. To find a successor to CEO John Akers, it appointed a committee that included Tom Murphy, CEO of ABC Cap Cities, and Jim Burke, the highly acclaimed former CEO of Johnson & Johnson (JNJ). The consensus of headhunters, pundits, security analysts, and media was that IBM’s new boss needed infotech experience, and some believed IBM especially needed someone who could win the PC war. Former Apple (AAPL) chief John Scully and Motorola’s George Fischer were highly touted candidates.
Murphy and Burke took a radically different view. They identified a few extremely specific criteria that they deemed central. In addition to the normal “givens,” such as character, integrity, values, and a strong record of making large-scale change, the duo decided on the following non-negotiable criteria: Customer orientation; business acumen – the ability to diagnose what ailed IBM and see how to fix it; and the ability to execute needed change and take the company forward.
Note that they didn’t use generic terms like “vision” or “strategy,” and they explicitly didn’t require infotech experience. The board’s choice was of course Lou Gerstner, CEO of RJR/Nabisco and previously president of American Express (AXP) — no techie, but outstanding on each of the three criteria. He famously made clear that IBM didn’t need a vision and then rescued the company from the verge of breakup, leaving it eight years later as one of the world’s most admired and valuable corporations.
Place the selection criteria into five buckets.
No platitudes or jargon allowed. The directors need criteria that lead to a specific, pointed conclusion.
Bucket 1: The “givens”: intelligence, character, a record of getting things done, a methodology for excellent execution, high energy. Two more are especially important: decisiveness, a bias toward saying “yes” or “no” rather than “maybe,” and courage. A dire lack of even one of these takes the candidate out of the running.
Bucket 2: Skills. Specific abilities like the ones Murphy and Burke identified for IBM’s CEO. They will vary widely from company to company and era to era.
MORE: Same-sex marriage: Will the Fortune 500 jump into the fray?
Bucket 3: Relationships. In most companies, much is achieved through a sustained network of external and internal relationships. It doesn’t happen fast. External relationships in particular are built over long periods and become valuable two-way bridges of information. What is the candidate’s record of building enduring relationships at several levels and converting them into strategic advantages?
Bucket 4: Judgment. Almost all decisions at the CEO level require tradeoffs, and many factors are qualitative and subjective. How effective is the candidate’s judgment on key questions, such as which information sources to listen to? In addition, how sage have the candidate’s judgments been on people, strategic bets, and resource allocation?
Bucket 5: Perception and cognition. Does a candidate see what waits around the bend before others do? Powers of perception and the ability to connect diverse external forces are worth a lot of points. What experiences signal that the candidate has this ability? What is his or her resilience in the face of shocks?
Confront common dilemmas.
Few candidates have it all, yet directors must choose someone. For example, here’s a common dilemma: choosing between a candidate possessing narrow, deep, highly sophisticated expertise but without much leadership experience, and a candidate with demonstrated leadership in a large organization but with no expertise in the sophisticated specialty urgently required for knowing where to take the company in the future.
That was the dilemma that Citigroup’s (C) board faced as it chose a replacement for CEO Chuck Prince in the financial crisis. To establish credibility and inspire confidence, most people were looking for the second type of candidate, someone with strong credentials as a successful leader from a large financial institution, and the chairman even made an offer to such a candidate. He was in essentially the same industry but without experience in toxic assets, which were at the heart of Citi’s crisis; he was successful in part because he had avoided such instruments. Through the heavy-handed persuasion of a highly regarded director who had lived with these toxic issues for much of his life, the board within hours reversed its position and turned to the first type of candidate, Vikram Pandit, the internal expert on toxic assets. The board believed Pandit would grow as a leader, and he did.
MORE: An Amazon-eBay throwdown is coming
Set the new CEO up for success.
No candidate is a perfect fit. The board should identify where the winning candidate falls short and determine how to help him or her close the gap. Directors need to discuss up front what risks they’re taking by choosing that candidate. They should also try to imagine how the candidate might evolve over time. An emerging best practice is to appoint a senior director as a sounding board and coach to help make the new CEO successful, with a highly structured plan for delivering that help in the CEO’s first year.
Corporate leadership has shifted from the CEO to the board. In this age of intense competition and accelerating change, boards must above all demonstrate excellence in their No. 1 job of having the right CEO at all times. Following these four practices will vastly improve their chances.
Is your board good enough to do the job they’re accountable for?