Deciding How to Decide

Definition

Leaders face a prior choice before every hard decision: not "what should we do?" but "how should we go about deciding?" Roberto names four managerial levers a leader controls to shape a decision process's quality — composition (who participates), context (what norms and ground rules govern deliberation), communication (how people exchange and evaluate ideas), and control (what role the leader plays and how much he directs versus opens up the process). Getting these right, before diving into content, raises the odds of both a well-reasoned decision and a plan people will actually implement.

In the Book

The chapter's spine is Kennedy's contrast between the 1961 Bay of Pigs invasion and the 1962 Cuban Missile Crisis — the same president, similar advisers, radically different processes. At the Bay of Pigs, Kennedy attended every meeting, participants spoke as departmental spokesmen deferring to rank and protocol, one small insulated subgroup drove the plan, and no one played devil's advocate — the group converged fast on a single flawed option. For the missile crisis, Kennedy redesigned all four levers: he deliberately stayed away from many preliminary meetings so people would speak freely; he told advisers to act as "skeptical generalists" examining the whole problem rather than defending their agency's turf; he split the group into two equal subgroups that developed rival plans (blockade vs. air strike) and exchanged critiques; and he assigned Robert Kennedy and Theodore Sorensen the explicit role of devil's advocates. Roberto also uses the Columbia shuttle disaster to show the cost of getting composition wrong — NASA's senior managers relied on job titles and protocol rather than reaching down to the engineers who had the relevant data and doubts.

Why It Matters

It separates two failure modes that get treated as one: a bad decision and a bad process. A leader can't fully control outcomes, which arrive only after the fact — but the process can be evaluated and corrected in real time, before the decision is locked in. Treating "who's in the room," "what norms apply," "how people talk to each other," and "how much the leader steers" as four independent design knobs — rather than an unexamined default — turns process design into something a leader does on purpose rather than something that just happens to them.