BETTER DECISIONS

IN THIS EDITION:

McKinsey Quarterly
1 The case for behavioral strategy
18 Untangling your organization’s decision making
30 A case study in combating bias
The case
for behavioral
strategy
Dan Lovallo and Olivier Sibony
 2

Left unchecked, subconscious biases will

undermine strategic decision making.
Here’s how to counter them and improve
corporate performance.

Once heretical, behavioral economics is now mainstream. Money

managers employ its insights about the limits of rationality
Dan Lovallo is
in understanding investor behavior and exploiting stock-pricing
a professor at the
anomalies. Policy makers use behavioral principles to boost
University of Sydney,
a senior research participation in retirement-savings plans. Marketers now understand why
fellow at the Institute some promotions entice consumers and others don’t.
for Business Innovation
at the University of Yet very few corporate strategists making important decisions
California, Berkeley, and
consciously take into account the cognitive biases—systematic tenden-
an adviser to McKinsey;
Olivier Sibony is a
cies to deviate from rational calculations—revealed by behavioral
director in McKinsey’s economics. It’s easy to see why: unlike in fields such as finance and
Brussels office. marketing, where executives can use psychology to make the most
3 March 2010

of the biases residing in others, in strategic decision making

leaders need to recognize their own biases. So despite growing aware-
ness of behavioral economics and numerous efforts by management
writers, including ourselves, to make the case for its application, most

its power.1

This is not to say that executives think their strategic decisions

are perfect. In a recent McKinsey Quarterly survey of 2,207 executives,
only 28 percent said that the quality of strategic decisions in their
companies was generally good, 60 percent thought that bad decisions
were about as frequent as good ones, and the remaining 12 percent
thought good decisions were altogether infrequent.2 Our candid conver-
sations with senior executives behind closed doors reveal a similar

body of research indicating that cognitive biases affect the most

important strategic decisions made by the smartest managers in the
best companies. Mergers routinely fail to deliver the expected
synergies.3 Strategic plans often ignore competitive responses.4 And
large investment projects are over budget and over time—over
and over again.5

In this article, we share the results of new research quantifying the

of this prize makes a strong case for practicing behavioral strategy—a

style of strategic decision making that incorporates the lessons of
psychology. It starts with the recognition that even if we try, like Baron
Münchhausen, to escape the swamp of biases by pulling ourselves up
by our own hair, we are unlikely to succeed. Instead, we need new norms
for activities such as managing meetings (for more on running
unbiased meetings, see “Taking the bias out of meetings” on
mckinsey.com/quarterly), gathering data, discussing analogies, and
stimulating debate that together can diminish the impact of
cognitive biases on critical decisions. To support those new norms,
we also need a simple language for recognizing and discussing biases, one
that is grounded in the reality of corporate life, as opposed to the

commitment and, in some organizations, a profound cultural change.

1
mckinsey.com/quarterly
and Dan P. Lovallo and Olivier Sibony, “Distortions and deceptions in strategic decisions,”
mckinsey.com/quarterly, February 2006.
2
See “Flaws in strategic decision making: McKinsey Global Survey Results,”
mckinsey.com/quarterly, January 2009.
3
See Dan Lovallo, Patrick Viguerie, Robert Uhlaner, and John Horn, “Deals without
delusions,” Harvard Business Review, December 2007, Volume 85, Number 12, pp. 92–99.
4
See John T. Horn, Dan P. Lovallo, and S. Patrick Viguerie, “Beating the odds in market
entry,” mckinsey.com/quarterly, November 2005.
5
See Bent Flyvbjerg, Dan Lovallo, and Massimo Garbuio, “Delusion and deception in large
infrastructure projects,” California Management Review, 2009, Volume 52, Number 1,
pp. 170–93.
Behavioral strategy print exhibits
Exhibit 1 of 2
Glance: The research analyzed a variety of decisions in areas that included
investments in new
Theproducts, M&A, strategy
case for behavioral and capital expenditures. 4

Exhibit title: About the research

What we did

1,048
Number of decisions analyzed

76%
Share of decisions related to M&A,
organizational change, or expansion
into new geographies, products,
and services

51%
Proportion of decisions that
could be attributed to a single,
specific business function
(sales, R&D, marketing,
manufacturing, or supply
chain/distribution)

The value of good decision processes

Think of a large business decision your company made recently: a major
acquisition, a large capital expenditure, a key technological choice,
or a new-product launch. Three things went into it. The decision almost
certainly involved some fact gathering and analysis. It relied on the
insights and judgment of a number of executives (a number sometimes
as small as one). And it was reached after a process—sometimes very
formal, sometimes completely informal—turned the data and judgment
into a decision.

Our research indicates that, contrary to what one might assume,

good analysis in the hands of managers who have good judgment won’t
naturally yield good decisions. The third ingredient—the process—
is also crucial. We discovered this by asking managers to report on both
the nature of an important decision and the process through which it
was reached. In all, we studied 1,048 major decisions made over the past
five years, including investments in new products, M&A decisions,
and large capital expenditures.
Glance:
industry Process
variables.has a greater effect on decision making than analysis or
industry
Exhibit variables.
title: Process carries weight
Exhibit title:
Process, Process
analysis, andcarries
industry weight
variables explain
decision-making effectiveness

Process,
Share analysis,
of performance and by
explained industry variables
given element explain
Process,
decision-making
(based analysis,
on multivariate and industry
effectiveness
regression analysis), % variables explain
decision-making effectiveness
Share of performance explained by givenQuantity element and detail of analysis
(based
Share on of multivariate
performance regression
explainedanalysis), %
by given performed—eg,
element detailed financial
(based on multivariate regression analysis), % modeling, sensitivity analysis, analysis of
financial reaction of markets
Quantity and detail of analysis
performed—eg,
Quantity detailed
and detail financial
of analysis
8
modeling, sensitivity analysis, analysis of
performed—eg, detailed financial
financial
modeling,reaction of markets
sensitivity analysis, analysisofofprocess to exploit
Quality
financial reaction of marketsanalysis and reach decision—eg,
Industry/company variables—eg, 8 explicit exploration of major uncertainties,
number of investment opportunities, 8 53
39 inclusion of perspectives that contradict
capital availability, predictability of Quality of process
senior leader’s point of to exploit
view, allowing
consumer tastes, availability of resources analysisofand
Quality reach to
process decision—eg,
exploit
Industry/company variables—eg, participation in discussion by skill and
to implement decision explicit
analysis exploration
and of major
reach uncertainties,
decision—eg,
number of investment opportunities, experience rather than by rank
Industry/company variables—eg, 53 inclusion of perspectives thatuncertainties,
contradict
capital availability, predictability of 39 explicit exploration of major
number of investment opportunities, 53 senior leader’s point of view, allowing
consumer tastes, availability of resources 39 inclusion of perspectives that contradict
capital availability, predictability of participation in point
discussion byallowing
skill and
to implement decision senior leader’s of view,
consumer tastes, availability of resources experience
participationrather than by rank
in discussion by skill and
to implement decision
experience rather than by rank

Note: To evaluate decision-making effectiveness, we asked respondents to assess

outcomes along four dimensions: revenue, profitability, market share, and productivity.

Note: To evaluate decision-making effectiveness, we asked respondents to assess

outcomes
Note: along four
To evaluate
Difference dimensions:
decision-making
in ROI between revenue,
top- profitability,
effectiveness, we asked
and bottom-quartile market share, and
respondents productivity.
to assess
outcomes along
decision fourpercentage
inputs, dimensions: revenue, profitability, market share, and productivity.
points

Difference in ROI between top- and bottom-quartile

decision in ROIpercentage
inputs,
Difference betweenpoints
top- and bottom-quartile
decision inputs, percentage points
others described the decision-making process: for instance, did you

Quality of process to exploit

6.9
analysis and reach decision
they have proved effective at overcoming biases.

Quantity
Quality ofand detailto
process ofexploit
analysis performed
5.3 6.9
and reach decision
Quality of process to exploit
we used and
analysis regression analysis to calculate how much of the variance 6.9
reach decision
in decision outcomes
Quantity and detail of
analysis
5.3
Quantity performed
and detail of
analysis performed
5.3

of people to believe that when their companies are successful or a decision turns out well,

mckinsey.com/quarterly,
the halo effect by asking respondents to focus on a typical decision process in their

market share, and productivity.

to implement decision
experience rather than by rank

The case for behavioral strategy 6

Note: To evaluate decision-making effectiveness, we asked respondents to assess

outcomes along four dimensions: revenue, profitability, market share, and productivity.

Difference in ROI between top- and bottom-quartile

decision inputs, percentage points

Quality of process to exploit

6.9
analysis and reach decision

Quantity and detail of

analysis performed
5.3

how much by the quantity and detail of the analysis. The answer:
process mattered more than analysis—by a factor of six. This
finding does not mean that analysis is unimportant, as a closer look
at the data reveals: almost no decisions in our sample made
through a very strong process were backed by very poor analysis.
Why? Because one of the things an unbiased decision-making process
will do is ferret out poor analysis. The reverse is not true; superb
analysis is useless unless the decision process gives it a fair hearing.

To get a sense of the value at stake, we also assessed the return on

investment (ROI) of decisions characterized by a superior process.8
The analysis revealed that raising a company’s game from the bottom
to the top quartile on the decision-making process improved its
ROI by 6.9 percentage points. The ROI advantage for top-quartile
versus bottom-quartile analytics was 5.3 percentage points, further
underscoring the tight relationship between process and analysis.
Good process, in short, isn’t just good hygiene; it’s good business.

8
This analysis covers the subset of 673 (out of all 1,048) decisions for which ROI data
were available.
7 March 2010

The building blocks of behavioral strategy

Any seasoned executive will of course recognize some biases and
take them into account. That is what we do when we apply a discount
factor to a plan from a direct report (correcting for that person’s
overoptimism). That is also what we do when we fear that one person’s
recommendation may be colored by self-interest and ask a neutral
third party for an independent opinion.

However, academic research and empirical observation suggest that

these corrections are too inexact and limited to be helpful. The
prevalence of biases in corporate decisions is partly a function of habit,
training, executive selection, and corporate culture. But most funda-
mentally, biases are pervasive because they are a product of human
nature—hard-wired and highly resistant to feedback, however brutal.
For example, drivers laid up in hospitals for traffic accidents they
themselves caused overestimate their driving abilities just as much
as the rest of us do.9

Improving strategic decision making therefore requires not only trying

to limit our own (and others’) biases but also orchestrating a decision-
making process that will confront different biases and limit their impact.
To use a judicial analogy, we cannot trust the judges or the jurors to
be infallible; they are, after all, human. But as citizens, we can expect
verdicts to be rendered by juries and trials to follow the rules of due
process. It is through teamwork, and the process that organizes it, that
we seek a high-quality outcome.

Building such a process for strategic decision making requires an under-

standing of the biases the process needs to address. In the discussion
that follows, we focus on the subset of biases we have found to be most
relevant for executives and classify those biases into five simple, business-
oriented groupings (for more on these groupings, see “A language to
discuss biases”). A familiarity with this classification is useful in itself
because, as the psychologist and Nobel laureate in economics Daniel
Kahneman has pointed out, the odds of defeating biases in a group
setting rise when discussion of them is widespread. But familiarity
alone isn’t enough to ensure unbiased decision making, so as we
discuss each family of bias, we also provide some general principles
and specific examples of practices that can help counteract it.

Counter pattern-recognition biases by changing

the angle of vision
The ability to identify patterns helps set humans apart but also carries
with it a risk of misinterpreting conceptual relationships. Common

9
Caroline E. Preston and Stanley Harris, “Psychology of drivers in traffic accidents,”
Journal of Applied Psychology, 1965, Volume 49, Number 4, pp. 284–88.
 The case for behavioral strategy 8

In most organizations, an executive

who projects great confidence in
a plan is more likely to get it approved
than one who lays out all the risks
and uncertainties surrounding it

pattern-recognition biases include saliency biases (which lead us to

overweight recent or highly memorable events) and the confirmation
bias (the tendency, once a hypothesis has been formed, to ignore
evidence that would disprove it). Particularly imperiled are senior
executives, whose deep experience boosts the odds that they will
rely on analogies, from their own experience, that may turn out to be
misleading.10 Whenever analogies, comparisons, or salient examples
are used to justify a decision, and whenever convincing champions use
their powers of persuasion to tell a compelling story, pattern-
recognition biases may be at work.

Pattern recognition is second nature to all of us—and often quite valuable—

so fighting biases associated with it is challenging. The best we
can do is to change the angle of vision by encouraging participants to
see facts in a different light and to test alternative hypotheses to
explain those facts. This practice starts with things as simple as field
and customer visits. It continues with meeting-management tech-
niques such as reframing or role reversal, which encourage participants
to formulate alternative explanations for the evidence with which
they are presented. It can also leverage tools, such as competitive war
games, that promote out-of-the-box thinking.

Sometimes, simply coaxing managers to articulate the experiences

influencing them is valuable. According to Kleiner Perkins partner
Randy Komisar, for example, a contentious discussion over manufac-
turing strategy at the start-up WebTV 11 suddenly became much more
manageable once it was clear that the preferences of executives about
which strategy to pursue stemmed from their previous career

10
For more on misleading experiences, see Sydney Finkelstein, Jo Whitehead, and Andrew
Campbell, Think Again: Why Good Leaders Make Bad Decisions and How to Keep It from
Happening to You, Boston: Harvard Business Press, 2008.
11
WebTV is now MSN TV.
9 March 2010

experience. When that realization came, he told us, there was

immediately a “sense of exhaling in the room.” Managers with software
experience were frightened about building hardware; managers
with hardware experience were afraid of ceding control to contract
manufacturers.

Getting these experiences into the open helped WebTV’s management

team become aware of the pattern recognition they triggered and
see more clearly the pros and cons of both options. Ultimately, WebTV’s
executives decided both to outsource hardware production to large
electronics makers and, heeding the worries of executives with hardware
experience, to establish a manufacturing line in Mexico as a backup,
in case the contractors did not deliver in time for the Christmas season.
That in fact happened, and the backup plan, which would not have
existed without a decision process that changed the angle of vision,
“saved the company.”

Another useful means of changing the angle of vision is to make it wider

by creating a reasonably large—in our experience at least six—set of
similar endeavors for comparative analysis. For example, in an effort
to improve US military effectiveness in Iraq in 2004, Colonel Kalev
Sepp—by himself, in 36 hours—developed a reference class of 53 similar
counterinsurgency conflicts, complete with strategies and outcomes.
This effort informed subsequent policy changes.12

Counter action-oriented biases by recognizing uncertainty

Most executives rightly feel a need to take action. However, the
actions we take are often prompted by excessive optimism about the
future and especially about our own ability to influence it. Ask
yourself how many plans you have reviewed that turned out to be based
on overly optimistic forecasts of market potential or underestimated
competitive responses. When you or your people feel—especially under
pressure—an urge to take action and an attractive plan presents itself,
chances are good that some elements of overconfidence have tainted it.

To make matters worse, the culture of many organizations suppresses

uncertainty and rewards behavior that ignores it. For instance, in
most organizations, an executive who projects great confidence in a plan
is more likely to get it approved than one who lays out all the risks and
uncertainties surrounding it. Seldom do we see confidence as a warning
sign—a hint that overconfidence, overoptimism, and other action-
oriented biases may be at work.

Superior decision-making processes counteract action-oriented biases

by promoting the recognition of uncertainty. For example, it often

12
Thomas E. Ricks, Fiasco: The American Military Adventure in Iraq, New York: Penguin
Press, 2006, pp. 393–94.
 The case for behavioral strategy 10

helps to make a clear and explicit distinction between decision

meetings, where leaders should embrace uncertainty while encouraging
dissent, and implementation meetings, where it’s time for executives
to move forward together. Also valuable are tools—such as scenario
planning, decision trees, and the “premortem” championed by research
psychologist Gary Klein (for more on the premortem, see “Strategic
decisions: When can you trust your gut?” on mckinsey.com/quarterly)—
that force consideration of many potential outcomes. And at the
time of a major decision, it’s critical to discuss which metrics need to
be monitored to highlight necessary course corrections quickly.

Counter stability biases by shaking things up

In contrast to action biases, stability biases make us less prone to
depart from the status quo than we should be. This category includes
anchoring—the powerful impact an initial idea or number has on
the subsequent strategic conversation. (For instance, last year’s numbers
are an implicit but extremely powerful anchor in any budget review.)
Stability biases also include loss aversion—the well-documented tendency
to feel losses more acutely than equivalent gains—and the sunk-
cost fallacy, which can lead companies to hold on to businesses they
should divest.13

One way of diagnosing your company’s susceptibility to stability biases

is to compare decisions over time. For example, try mapping the
percentage of total new investment each division of the company receives
year after year. If that percentage is stable but the divisions’ growth

a common one. Our research indicates, for example, that in multi-

business corporations over a 15-year time horizon, there is a near-perfect
correlation between a business unit’s current share of the capital
expenditure budget and its budget share in the previous year. A similar
inertia often bedevils advertising budgets and R&D project pipelines.

One way to help managers shake things up is to establish stretch targets

that are impossible to achieve through “business as usual.” Zero-
based (or clean-sheet) budgeting sounds promising, but in our experience
companies use this approach only when they are in dire straits. An

percentage (for instance, 10 percent). The resulting tough choices

facilitate the redeployment of resources to more valuable opportunities.
Finally, challenging budget allocations at a more granular level can
help companies reprioritize their investments.14

13
See John T. Horn, Dan P. Lovallo, and S. Patrick Viguerie, “Learning to let go: Making
better exit decisions,” mckinsey.com/quarterly, May 2006.
14
For more on reviewing the growth opportunities available across different micromarkets
ranging in size from $50 million to $200 million, rather than across business units
as a whole, see Mehrdad Baghai, Sven Smit, and Patrick Viguerie, “Is your growth strategy
Harvard Business Review, May 2009, Volume 87, Number 5, pp. 86–96.
11 March 2010

Counter interest biases by making them explicit

Misaligned incentives are a major source of bias. “Silo thinking,” in
which organizational units defend their own interests, is its most easily
detectable manifestation. Furthermore, senior executives sometimes
honestly view the goals of a company differently because of their different
roles or functional expertise. Heated discussions in which participants

the presence of different (and generally unspoken) interest biases.

of “interests,” including reputation, career options, and individual

preferences, leads to the inescapable conclusion that there will always

managers and the company as a whole. Strong decision-making

processes explicitly account for diverging interests. For example, if
before the time of a decision, strategists formulate precisely the criteria

for individual managers to change the terms of the debate to make their
preferred actions seem more attractive. Similarly, populating
meetings or teams with participants whose interests clash can reduce
the likelihood that one set of interests will undermine thoughtful
decision making.

Counter social biases by depersonalizing debate

Social biases are sometimes interpreted as corporate politics but in fact
are deep-rooted human tendencies. Even when nothing is at stake,
we tend to conform to the dominant views of the group we belong to (and
of its leader).15 Many organizations compound these tendencies
because of both strong corporate cultures and incentives to conform.
An absence of dissent is a strong warning sign. Social biases also
are likely to prevail in discussions where everyone in the room knows
the views of the ultimate decision maker (and assumes that the
leader is unlikely to change her mind).

Countless techniques exist to stimulate debate among executive teams,

and many are simple to learn and practice. (For more on promoting
debate, see suggestions from Kleiner Perkins’ Randy Komisar and Xerox’s

decision making” on mckinsey.com/quarterly.) But tools per se won’t

create debate: that is a matter of behavior. Genuine debate requires
diversity in the backgrounds and personalities of the decision makers, a
climate of trust, and a culture in which discussions are depersonalized.

15
The Asch conformity experiments, conducted during the 1950s, are a classic example
of this dynamic. In the experiments, individuals gave clearly incorrect answers to
simple questions after confederates of the experimenter gave the same incorrect answers
aloud. See Solomon E. Asch, “Opinions and social pressure,” 1955,
Volume 193, Number 5, pp. 31–35.
 The case for behavioral strategy 12

Populating meetings or teams with

participants whose interests clash can
reduce the likelihood that one set
of interests will undermine thoughtful
decision making

Most crucially, debate calls for senior leaders who genuinely believe
in the collective intelligence of a high-caliber management team. Such
executives see themselves serving not only as the ultimate decision
makers but also as the orchestrators of disciplined decision processes.
They shape management teams with the humility to encourage

debate without damaging personal relationships. We do not suggest that

CEOs should become humble listeners who rely solely on the consensus
of their teams—that would substitute one simplistic stereotype for
another. But we do believe that behavioral strategy will founder without
their leadership and role modeling.

Four steps to adopting behavioral strategy

Our readers will probably recognize some of these ideas and tools
as techniques they have used in the past. But techniques by themselves
will not improve the quality of decisions. Nothing is easier, after
all, than orchestrating a perfunctory debate to justify a decision already
made (or thought to be made) by the CEO. Leaders who want to shape
the decision-making style of their companies must commit themselves
to a new path.

1 Some executives fear that applying the principles we describe here

could be divisive, counterproductive, or simply too time consuming (for
more on the dangers of decision paralysis, see the commentary by
Decide which
decisions on strategic decision making” on mckinsey.com/quarterly). We share
warrant the this concern and do not suggest applying these principles to all
effort decisions. Here again, the judicial analogy is instructive. Just as higher
standards of process apply in a capital case than in a proceeding before
a small-claims court, companies can and should pay special attention to
two types of decisions.
13 March 2010

The first set consists of rare, one-of-a-kind strategic decisions. Major

mergers and acquisitions, “bet the company” investments, and crucial
technological choices fall in this category. In most companies, these
decisions are made by a small subgroup of the executive team, using an
ad hoc, informal, and often iterative process. The second set includes
repetitive but high-stakes decisions that shape a company’s strategy over
time. In most companies, there are generally no more than one or
two such crucial processes, such as R&D allocations in a pharmaceutical
company, investment decisions in a private-equity firm, or capital
expenditure decisions in a utility. Formal processes—often affected by
biases—are typically in place to make these decisions.

2 Open discussion of the biases that may be undermining decision making

is invaluable. It can be stimulated both by conducting postmortems
of past decisions and by observing current decision processes. Are we
Identify at risk, in this meeting, of being too action oriented? Do I see someone
the biases who thinks he recognizes a pattern but whose choice of analogies seems
most likely to
misleading to me? Are we seeing biases combine to create dysfunc-
affect critical
tional patterns that, when repeated in an organization, can become
decisions
cultural traits? For example, is the combination of social and status quo
biases creating a culture of consensus-based inertia? This discussion
will help surface the biases to which the decision process under review
is particularly prone.

3 Companies should select mechanisms that are appropriate to the

type of decision at hand, to their culture, and to the decision-making
styles of their leaders. For instance, one company we know counters
Select
social biases by organizing, as part of its annual planning cycle,
practices and
a systematic challenge by outsiders to its business units’ plans. Another
tools to
fights pattern-recognition biases by asking managers who present
counter the
most relevant a recommendation to share the raw data supporting it, so other
biases executives in this analytically minded company can try to discern
alternative patterns.

If, as you read these lines, you have already thought of three reasons
these techniques won’t work in your own company’s culture, you
are probably right. The question is which ones will. Adopting behavioral
strategy means not only embracing the broad principles set forth
above but also selecting and tailoring specific debiasing practices to
turn the principles into action.
 The case for behavioral strategy 14

4Embed
By embedding these practices in formal corporate operating procedures
(such as capital-investment approval processes or R&D reviews),
executives can ensure that such techniques are used with some regularity
and not just when the ultimate decision maker feels unusually
practices in uncertain about which call to make. One reason it’s important to embed
formal these practices in recurring procedures is that everything we know
processes
about the tendency toward overconfidence suggests that it is unwise to
rely on one’s instincts to decide when to rely on one’s instincts!
Another is that good decision making requires practice as a management
team: without regular opportunities, the team will agree in principle
on the techniques it should use but lack the experience (and the mutual
trust) to use them effectively.

The behavioral-strategy journey requires effort and the commitment

of senior leadership, but the payoff—better decisions, not to
mention more engaged managers—makes it one of the most valuable
strategic investments organizations can make.

Copyright © 2010 McKinsey & Company. All rights reserved.

We welcome your comments on this article. Please send them to
[email protected].
 A language
to discuss biases
Psychologists and behavioral economists have identified dozens of cognitive
biases. The typology we present here is not meant to be exhaustive but rather
to focus on those biases that occur most frequently and that have the largest
impact on business decisions. As these groupings make clear, one of the insidious
things about cognitive biases is their close relationship with the rules of thumb
and mind-sets that often serve managers well. For example, many a seasoned
executive rightly prides herself on pattern-recognition skills cultivated over
This bias typology was the years. Similarly, seeking consensus when making a decision is often not a
prepared by Dan Lovallo failing but a condition of success. And valuing stability rather than “rocking
and Olivier Sibony. the boat” or “fixing what ain’t broke” is a sound management precept.

Action-oriented biases
drive us to take action less thoughtfully than we should.

Excessive optimism. The tendency Overconfidence. Overestimating

for people to be overoptimistic our skill level relative to others’, leading
about the outcome of planned actions, us to overestimate our ability to
to overestimate the likelihood of affect future outcomes, take credit for
positive events, and to underestimate past outcomes, and neglect the role
the likelihood of negative ones. of chance.

Competitor neglect. The tendency

to plan without factoring in competi-
tive responses, as if one is playing tennis
against a wall, not a live opponent.

Interest biases
arise in the presence of conflicting incentives, including nonmonetary
and even purely emotional ones.

Misaligned individual Inappropriate attachments.

incentives. Incentives for individuals Emotional attachment of individuals
in organizations to adopt views or to people or elements of the business
to seek outcomes favorable to their unit (such as legacy products or brands),
or themselves, at the expense of creating a misalignment of interests.1
the overall interest of the company.
These self-serving views are often held Misaligned perception of
genuinely, not cynically. corporate goals. Disagreements
(often unspoken) about the hierarchy or
relative weight of objectives pursued
by the organization and about the trade-
offs between them.

1
Sydney Finkelstein, Jo Whitehead, and Andrew Campbell, Think Again: Why Good
Leaders Make Bad Decisions and How to Keep It fromHappening to You, Boston: Harvard
Business Press, 2008.
 Pattern-recognition biases
lead us to recognize patterns even where there are none.

Confirmation bias. The over- Power of storytelling. The

weighting of evidence consistent with tendency to remember and to believe
a favored belief, underweighting more easily a set of facts when they
of evidence against a favored belief, are presented as part of a coherent
or failure to search impartially for story.
evidence.
Champion bias. The tendency
Management by example. to evaluate a plan or proposal based
Generalizing based on examples that on the track record of the person
are particularly recent or memorable. presenting it, more than on the facts
supporting it.
False analogies—especially,
misleading experiences.
Relying on comparisons with situations
that are not directly comparable.

Stability biases
create a tendency toward inertia in the presence of uncertainty.

Anchoring and insufficient Sunk-cost fallacy. Paying

adjustment. Rooting oneself to an attention to historical costs that are
initial value, leading to insufficient not recoverable when considering
adjustments of subsequent estimates. future courses of action.

Loss aversion. The tendency to feel Status quo bias. Preference

losses more acutely than gains of for the status quo in the absence of
the same amount, making us more risk- pressure to change it.
averse than a rational calculation
would suggest.

Social biases
arise from the preference for harmony over conflict.

Groupthink. Striving for consensus Sunflower management.

at the cost of a realistic appraisal of Tendency for groups to align with
alternative courses of action. the views of their leaders,
whether expressed or assumed.

To listen to the authors narrate a more comprehensive presentation of

these biases and the ways they can combine to create dysfunctional patterns
in corporate cultures, visit mckinsey.com/quarterly.
 June 2017

Untangling your
organization’s
decision making
Any organization can improve the speed and quality of its decisions
by paying more attention to what it’s deciding.

by Aaron De Smet, Gerald Lackey, and Leigh M. Weiss

It’s the best and worst of times for decision makers. Swelling stockpiles
of data, advanced analytics, and intelligent algorithms are providing
organizations with powerful new inputs and methods for making all manner
of decisions. Corporate leaders also are much more aware today than
they were 20 years ago of the cognitive biases—anchoring, loss aversion,
confirmation bias, and many more—that undermine decision making without
our knowing it. Some have already created formal processes—checklists,
devil’s advocates, competing analytic teams, and the like—to shake up the
debate and create healthier decision-making dynamics.

Now for the bad news. In many large global companies, growing
organizational complexity, anchored in strong product, functional, and
regional axes, has clouded accountabilities. That means leaders are less
able to delegate decisions cleanly, and the number of decision makers has
risen. The reduced cost of communications brought on by the digital age has
compounded matters by bringing more people into the flow via email, Slack,
and internal knowledge-sharing platforms, without clarifying decision-
making authority. The result is too many meetings and email threads with
too little high-quality dialogue as executives ricochet between boredom and
disengagement, paralysis, and anxiety (Exhibit 1). All this is a recipe for poor
decisions: 72 percent of senior-executive respondents to a McKinsey survey
said they thought bad strategic decisions either were about as frequent as good
ones or were the prevailing norm in their organization.

18
 QWeb 2017
Decision making
Exhibit 1 of 7
Exhibit 1

Growing organizational complexity and proliferating digital communications

are a recipe for poor decisions.

Disengagement Paralysis Anxiety

Hearing a presentation for Stymied by too much data The stakes are too high
the hundredth time

The ultimate solution for many organizations looking to untangle their

decision making is to become flatter and more agile, with decision authority
and accountability going hand in hand. High-flying technology companies
such as Google and Spotify are frequently the poster children for this
approach, but it has also been adapted by more traditional ones such as
ING (for more, see our recent McKinsey Quarterly interview “ING’s agile
transformation,” on McKinsey.com). As we’ve described elsewhere,1 agile
organization models get decision making into the right hands, are faster in
reacting to (or anticipating) shifts in the business environment, and often
become magnets for top talent, who prefer working at companies with fewer
layers of management and greater empowerment.

As we’ve worked with organizations seeking to become more agile, we’ve

found that it’s possible to accelerate the improvement of decision making
through the simple steps of categorizing the type of decision that’s being
made and tailoring your approach accordingly. In our work, we’ve observed
four types of decisions (Exhibit 2):

1
 ee Wouter Aghina, Aaron De Smet, and Kirsten Weerda, “Agility: It rhymes with stability,” McKinsey Quarterly,
S
December 2015, McKinsey.com.

19
 QWeb 2017
Decision making
Exhibit 2 of 7
Exhibit 2

The ABCDs of categorizing decisions.

Big-bet decisions with major Cross-cutting decisions

Broad

consequences for the company, that are frequent and require

often involving situations with broad collaboration across
unclear right or wrong choices organizational boundaries
Scope
and
impact
Ad hoc decisions that arise D elegated decisions that
Narrow

episodically; impact on broader can be assigned to individual

organization depends upon how primarily accountable or to
concentrated they are working team

Unfamiliar, infrequent Familiar, frequent

Level of familiarity

• Big-bet decisions. These infrequent and high-risk decisions have the

potential to shape the future of the company.

• Cross-cutting decisions. In these frequent and high-risk decisions, a series

of small, interconnected decisions are made by different groups as part of a
collaborative, end-to-end decision process.

•D
 elegated decisions. These frequent and low-risk decisions are effectively
handled by an individual or working team, with limited input from others.

• Ad hoc decisions. The organization’s infrequent, low-stakes decisions

are deliberately ignored in this article, in order to sharpen our focus on
the other three areas, where organizational ambiguity is most likely to
undermine decision-making effectiveness.

These decision categories often get overlooked, in our experience, because

organizational complexity, murky accountabilities, and information
overload have conspired to create messy decision-making processes in
many companies. In this article, we’ll describe how to vary your decision-
making methods according to the circumstances. We’ll also offer some tools
that individuals can use to pinpoint problems in the moment and to take
corrective action that should improve both the decision in question and, over
time, the organization’s decision-making norms.

Before we begin, we should emphasize that even though the examples

we describe focus on enterprise-level decisions, the application of this
framework will depend on the reader’s perspective and location in the

20
 organization. For example, what might be a delegated decision for the
enterprise as a whole could be a big-bet decision for an individual business
unit. Regardless, any fundamental change in decision-making culture needs
to involve the senior leaders in the organization or business unit. The top
team will decide what decisions are big bets, where to appoint process leaders
for cross-cutting decisions, and to whom to delegate. Senior executives also
serve the critical functions of role-modeling a culture of collaboration and of
making sure junior leaders take ownership of the delegated decisions.

BIG BETS
Bet-the-company decisions—from major acquisitions to game-changing
capital investments—are inherently the most risky. Efforts to mitigate the
impact of cognitive biases on decision making have, rightly, often focused
on big bets. And that’s not the only special attention big bets need. In our
experience, steps such as these are invaluable for big bets:

• Appoint an executive sponsor. Each initiative should have a sponsor, who

will work with a project lead to frame the important decisions for senior
leaders to weigh in on—starting with a clear, one-sentence problem
statement.

• Break things down, and connect them up. Large, complex decisions often
have multiple parts; you should explicitly break them down into bite-size
chunks, with decision meetings at each stage. Big bets also frequently
have interdependencies with other decisions. To avoid unintended
consequences, step back to connect the dots.

• Deploy a standard decision-making approach. The most important way

to get big-bet decisions right is to have the right kind of interaction and
discussion, including quality debate, competing scenarios, and devil’s
advocates. Critical requirements are to create a clear agenda that focuses
on debating the solution (instead of endlessly elaborating the problem),
to require robust prework, and to assemble the right people, with diverse
perspectives.

• Move faster without losing commitment. Fast-but-good decision making

also requires bringing the available facts to the table and committing to
the outcome of the decision. Executives have to get comfortable living
with imperfect data and being clear about what “good enough” looks like.
Then, once a decision is made, they have to be willing to commit to it and
take a gamble, even if they were opposed during the debate. Make sure, at
the conclusion of every meeting, that it is clear who will communicate the
decision and who owns the actions to begin carrying it out.

21
 An example of a company that does much of this really well is a
semiconductor company that believes so much in the importance of getting
big bets right that it built a whole management system around decision
making. The company never has more than one person accountable for
decisions, and it has a standard set of facts that need to be brought into
any meeting where a decision is to be made (such as a problem statement,
recommendation, net present value, risks, and alternatives). If this
information isn’t provided, then a discussion is not even entertained. The
CEO leads by example, and to date, the company has a very good track record
of investment performance and industry-changing moves.

It’s also important to develop tracking and feedback mechanisms to judge the
success of decisions and, as needed, to course correct for both the decision
and the decision-making process. One technique a regional energy provider
uses is to create a one-page self-evaluation tool that allows each member of
the team to assess how effectively decisions are being made and how well
the team is adhering to its norms. Members of key decision-making bodies
complete such evaluations at regular intervals (after every fifth or tenth
meeting). Decision makers also agree, before leaving a meeting where a
decision has been made, how they will track project success, and they set a
follow-up date to review progress against expectations.

Big-bet decisions often are easy to recognize, but not always (Exhibit 3).
Sometimes a series of decisions that might appear small in isolation
represent a big bet when taken as a whole. A global technology company we
know missed several opportunities that it could have seized through big-bet
investments, because it was making technology-development decisions
independently across each of its product lines, which reduced its ability to
recognize far-reaching shifts in the industry. The solution can be as simple
as a mechanism for periodically categorizing important decisions that are
being made across the organization, looking for patterns, and then deciding
whether it’s worthwhile to convene a big-bet-style process with executive
sponsorship. None of this is possible, though, if companies aren’t in the habit
of isolating major bets and paying them special attention.

CROSS-CUTTING DECISIONS
Far more frequent than big-bet decisions are cross-cutting ones—think
pricing, sales, and operations planning processes or new-product launches—
that demand input from a wide range of constituents. Collaborative efforts
such as these are not actually single-point decisions, but instead comprise
a series of decisions made over time by different groups as part of an end-
to-end process. The challenge is not the decisions themselves but rather

22
 QWeb 2017
Decision making
Exhibit 3 of 7
Exhibit 3
A belated heads-up means you are not recognizing big bets.

The problem: Missing your “Bs” (big bets)

Symptoms Example
Senior leaders are surprised when Wealth-management company where
they hear about the decision business-unit leaders made significant,
independent commitments of capital
Decision has big implications for the
in M&A decisions, constraining options
organization, but some relevant senior
for rest of business
leaders are not in the room

Fixing the problem

Questions to ask Mind-set to overcome

What are the implications for the “I can make any decision that affects
organization? my part of the business”
Would someone higher up want to have
input into this decision?

the choreography needed to bring multiple parties together to provide the

right input, at the right time, without breeding bureaucracy that slows down
the process and can diminish the decision quality. This is why the common
advice to focus on “who has the decision” (or, “the D”) isn’t the right starting
point; you should worry more about where the key points of collaboration
and coordination are.

It’s easy to err by having too little or too much choreography. For an example
of the former, consider the global pension fund that found itself in a major
cash crunch because of uncoordinated decision making and limited
transparency across its various business units. A perfect storm erupted when
different business units’ decisions simultaneously increased the demand for
cash while reducing its supply. In contrast, a specialty-chemicals company
experienced the pain of excess choreography when it opened membership
on each of its six governance committees to all senior leaders without
clarifying the actual decision makers. All participants felt they had a right
(and the need) to express an opinion on everything, even where they had
little knowledge or expertise. The purpose of the meetings morphed into
information sharing and unstructured debate, which stymied productive
action (Exhibit 4).

23
 QWeb 2017
Decision making
Exhibit 4 of 7
Exhibit 4
Too many cooks get involved in the absence of processes
for cross-cutting decisions.

The problem: Treating a “C” (cross-cutting decision) as a “B” (big bet)

Symptoms Example
Decisions have major implications for parts Specialty-chemicals company where
of business whose stakeholders aren’t every R&D stage-gate decision went
involved, resulting in poor decisions to executive team for review, though
the team lacked the expertise to make
Important decisions get slowed down by
a reasoned call
largely unnecessary committee meetings
and approvals

Fixing the problem

Questions to ask Mind-set to overcome

Are we making this same type of decision “This is an important decision that can’t be
on a regular basis? made without senior-most approval, even
though we make these decisions regularly”
Do we have the relevant stakeholders with
expertise to inform the decision involved?

Whichever end of the spectrum a company is on with cross-cutting decisions,

the solution is likely to be similar: defining roles and decision rights along
each step of the process. That’s what the specialty-chemicals company did.
Similarly, the pension fund identified its CFO as the key decision maker in
a host of cash-focused decisions, and then it mapped out the decision rights
and steps in each of the contributing processes. For most companies seeking
enhanced coordination, priorities include:

•Map out the decision-making process, and then pressure-test it. Identify
decisions that involve a cross-cutting group of leaders, and work with
the stakeholders of each to agree on what the main steps in the process
entail. Lay out a simple, plain-English playbook for the process to define
the calendar, cadence, handoffs, and decisions. Too often, companies find
themselves building complex process diagrams that are rarely read or used
beyond the team that created them. Keep it simple.

•Run water through the pipes. Then work through a set of real-life scenarios
to pressure-test the system in collaboration with the people who will
be running the process. We call this process “running water through
the pipes,” because the first several times you do it, you will find where
the “leaks” are. Then you can improve the process, train people to work

24
 within (and, when necessary, around) it, and confront, when the stakes are
relatively low, leadership tensions or stresses in organizational dynamics.

•Establish governance and decision-making bodies. Limit the number

of decision-making bodies, and clarify for each its mandate, standing
membership, roles (decision makers or critical “informers”), decision-
making protocols, key points of collaboration, and standing agenda.
Emphasize to the members that committees are not meetings but decision-
making bodies, and they can make decisions outside of their standard
meeting times. Encourage them to be flexible about when and where they
make decisions, and to focus always on accelerating action.

• Create shared objectives, metrics, and collaboration targets. These will

help the persons involved feel responsible not just for their individual
contributions in the process, but also for the process’s overall effectiveness.
Team members should be encouraged to regularly seek improvements in
the underlying process that is giving rise to their decisions.

Getting effective at cross-cutting decision making can be a great way to

tackle other organizational problems, such as siloed working (Exhibit 5).
Take, for example, a global finance company with a matrix of operations
across markets and regions that struggled with cross-business-unit decision
QWeb 2017
Decision making
Exhibit 5 of 7
Exhibit 5
When you are locked in silos, you are unlikely to collaborate
effectively on cross-cutting decisions.

The problem: Treating a “C” (cross-cutting decision) as a “D” (delegated)

Symptoms Example
Decisions create value for 1 part of Financial company where 1 business
business at the expense of others or unit changed its product without
the entire enterprise considering impact on profit and
loss for other product business units
Executives feel they don’t know the
organization-wide strategy or what
different parts of business are doing

Fixing the problem

Questions to ask Mind-set to overcome

Who are the stakeholders in this decision? “My obligation is to my part of the
organization, not the enterprise as a whole”
How do we facilitate an open and rapid
flow of information?

25
 making. Product launches often cannibalized the products of other market
groups. When the revenue shifts associated with one such decision caught
the attention of senior management, company leaders formalized a new
council for senior executives to come together and make several types of
cross-cutting decisions, which yielded significant benefits.

DELEGATED DECISIONS
Delegated decisions are far narrower in scope than big-bet decisions or cross-
cutting ones. They are frequent and relatively routine elements of day-to-day
management, typically in areas such as hiring, marketing, and purchasing.
The value at stake for delegated decisions is in the multiplier effect they can
have because of the frequency of their occurrence across the organization.
Placing the responsibility for these decisions in the hands of those closest
to the work typically delivers faster, better, and more efficiently executed
decisions, while also enhancing engagement and accountability at all levels
of the organization.

In today’s world, there is the added complexity that many decisions (or
parts of them) can be “delegated” to smart algorithms enabled by artificial
intelligence. Identifying the parts of your decisions that can be entrusted to
intelligent machines will speed up decisions and create greater consistency
and transparency, but it requires setting clear thresholds for when those
systems should escalate to a person, as well as being clear with people about
how to leverage the tools effectively.

It’s essential to establish clarity around roles and responsibilities in order to

craft a smooth-running system of delegated decision making (Exhibit 6).
A renewable-energy company we know took this task seriously when
undergoing a major reorganization that streamlined its senior management
and drove decisions further down in the organization. The company
developed a 30-minute “role card” conversation for each manager to
have with his or her direct reports. As part of this conversation, managers
explicitly laid out the decision rights and accountability metrics for each
direct report. This approach allowed the company’s leaders to decentralize
their decision making while also ensuring that accountability and
transparency were in place. Such role clarity enables easier navigation,
speeds up decision making, and makes it more customer focused. Companies
may find it useful to take some of the following steps to reorganize decision-
making power and establish transparency in their organization:

26
 QWeb 2017
Decision making
Exhibit 6 of 7
Exhibit 6
Drawn-out and complicated processes often mean more
delegating is needed.

The problem: Treating a “D” (delegated decision) as a “C” (cross-cutting)

Symptoms Example
Decisions that should be quick Energy company where changes to
seem to take forever and involve HR or finance policies were governed
more alignment than needed by executive committee instead of
delegated to head of HR or CFO
Decisions become unnecessarily
complex because of efforts to
incorporate all stakeholder input

Fixing the problem

Questions to ask Mind-set to overcome

Is there a single role that could “Delegating is risky; we don’t just let
make this decision (eg, it’s part of the people collect input from others and
job description)? then decide whatever they want”
Who needs to provide input but has
no “vote”?

• Delegate more decisions. To start delegating decisions today, make a list

of the top 20 regularly occurring decisions. Take the first decision and ask
three questions: (1) Is this a reversible decision? (2) Does one of my direct
reports have the capability to make this decision? (3) Can I hold that person
accountable for making the decision? If the answer to these questions is
yes, then delegate the decision. Continue down your list of decisions until
you are only making decisions for which there is one shot to get it right and
you alone possess the capabilities or accountability. The role-modeling of
senior leaders is invaluable, but they may be reluctant. Reassure them (and
yourself) by creating transparency through good performance dashboards,
scorecards, and key performance indicators (KPIs), and by linking metrics
back to individual performance reviews.

• Avoid overlap of decision rights. Doubling up decision responsibility across

management levels or dimensions of the reporting matrix only leads to
confusion and stalemates. Employees perform better when they have
explicit authority and receive the necessary training to tackle problems on
their own. Although it may feel awkward, leaders should be explicit with
their teams about when decisions are being fully delegated and when the
leaders want input but need to maintain final decision rights.

27
 • Establish a clear escalation path. Set thresholds for decisions that require
approval (for example, spending above a certain amount), and lay out a
specific protocol for the rare occasion when a decision must be kicked up
the ladder. This helps mitigate risk and keeps things moving briskly.

• Don’t let people abdicate. One of the key challenges in delegating decisions
is actually getting people to take ownership of the decisions. People will
often succumb to escalating decisions to avoid personal risk; leaders
need to play a strong role in encouraging personal ownership, even (and
especially) when a bad call is made.

This last point deserves elaboration: although greater efficiency comes

with delegated decision making, companies can never completely eliminate
mistakes, and it’s inevitable that a decision here or there will end badly. What
executives must avoid in this situation is succumbing to the temptation to
yank back control (Exhibit 7). One CEO at a Fortune 100 company learned
this lesson the hard way. For many years, her company had worked under
a decentralized decision-making framework where business-unit leaders
could sign off on many large and small deals, including M&A. Financial
underperformance and the looming risk of going out of business during a
severe market downturn led the CEO to pull back control and centralize
virtually all decision making. The result was better cost control at the
expense of swift decision making. After several big M&A deals came and
QWeb 2017
Decision making
Exhibit 7 of 7
Exhibit 7
Top-heavy processes often mean more delegating is needed.

The problem: Treating a “D” (delegated decision) as a “B” (big bet)

Symptoms Example
Senior executives (want to) control High-tech company that required CEO
decisions that should rightfully be to sign off on all new hires at any level of
made lower in the organization the organization
Escalation of decisions to top of
organization is common

Fixing the problem

Questions to ask Mind-sets to overcome

What is the lowest level of accountability “I need to be involved in all decisions”
at which this decision could be made? (senior executive)

What skills and capabilities are needed “I can’t make a decision on my own,
to make this decision? because that’s not how we do things here”

28
 went because the organization was too slow to act, the CEO decided she had
to decentralize decisions again. This time, she reinforced the decentralized
system with greater leadership accountability and transparency.

Instead of pulling back decision power after a slipup, hold people accountable
for the decision, and coach them to avoid repeating the misstep. Similarly,
in all but the rarest of cases, leaders should resist weighing in on a decision
kicked up to them during a logjam. From the start, senior leaders should
collectively agree on escalation protocols and stick with them to create
consistency throughout the organization. This means, when necessary, that
leaders must vigilantly reinforce the structure by sending decisions back
with clear guidance on where the leader expects the decision to be made
and by whom. If signs of congestion or dysfunction appear, leaders should
reexamine the decision-making structure to make sure alignment, processes,
and accountability are optimally arranged.

None of this is rocket science. Indeed, the first decision-making step Peter
Drucker advanced in “The effective decision,” a 1967 Harvard Business
Review article, was “classifying the problem.” Yet we’re struck, again and
again, by how few large organizations have simple systems in place to make
sure decisions are categorized so that they can be made by the right people
in the right way at the right time. Interestingly, Drucker’s classification
system focused on how generic or exceptional the problem was, as opposed
to questions about the decision’s magnitude, potential for delegation, or
cross-cutting nature. That’s not because Drucker was blind to these issues;
in other writing, he strongly advocated decentralizing and delegating
decision making to the degree possible. We’d argue, though, that today’s
organizational complexity and rapid-fire digital communications have
created considerably more ambiguity about decision-making authority than
was prevalent 50 years ago. Organizations haven’t kept up. That’s why the
path to better decision making need not be long and complicated. It’s simply a
matter of untangling the crossed web of accountability, one decision at a time.

Aaron De Smet is a senior partner in McKinsey’s Houston office, Gerald Lackey is an

expert in the Washington, DC, office, and Leigh Weiss is a senior expert in the Boston office.

Copyright © 2017 McKinsey & Company. All rights reserved.

29
 May 2017

A case study in combating bias

Following several disappointing investments, the German electric
utility RWE overhauled its decision-making processes. Learn how
from the CFO who spearheaded the effort.

The Quarterly: Tell us a bit about the circumstances that motivated RWE’s
management to undertake a broad debiasing operation.

Bernhard Günther: In the second half of the last decade, we spent more
than €10 billion on big capital-expenditure programs and acquisitions in
conventional power plants. In the business cases underlying these decisions,
we were betting on the assumptions of ever-rising commodity prices, ever-
rising power prices. We were not alone in our industry in hitting a kind of
investment peak at that time. What we and most other peers totally under-
estimated was the turnaround in public sentiment toward conventional power
generation—for example, the green transformation of the German energy
system, and the technological progress in renewable generation and related
production costs. These factors went in a completely opposite direction
compared to our scenarios.

Conventional power generation in continental Europe went through the

deepest crisis the industry has ever seen. This ultimately led to the split of the
two biggest German players in the industry, E.ON and RWE. Both companies
separated their ailing conventional power-generation businesses from the rest
of the company.

The Quarterly: Was it difficult to convince members of the executive and

supervisory boards to scrutinize your decision-making practices?

Bernhard Günther: Actually, it was the supervisory board asking, “Where has
the shareholders’ money gone?” and we in the executive board wanted to learn
our lessons from this experience as well. So we embarked on a postmortem
analysis to understand what went wrong and why, by looking at a sample of
these €10 billion investments. We asked ourselves, “Is there anything we could

30
 have done differently, and if so, how can we learn from this in the future?”
The spirit of it was not about shaming and blaming, but about learning from
our own mistakes.

The Quarterly: What were the main contributing factors that you identified in
your investigation?

Bernhard Günther: There were a few outright areas of managerial under-

performance such as some time and cost overruns on the €10 billion
investments, totally unrelated to external factors. There were also exogenous
factors that were not in our base-case assumption but that should have
been within our solution space—the most obvious being the political intent to
push renewables into the market, which was publicly known at the time our
investment decisions were made. There was also at least one unforeseeable
factor—the Fukushima disaster. The German government reacted by
rushing into a sudden exit from nuclear-power generation. Roughly half of
the nuclear plants were switched off immediately, significantly shortening
the economic lifetime of the remaining plants. But even if you discount
for Fukushima, I think the ultimate end game wouldn’t have looked much
different from today’s perspective; it just speeded the whole thing up.

The Quarterly: As you analyzed the decision-making dynamics at work, what

biases did you start to see?

Bernhard Günther: What became obvious is that we had fallen victim to a

number of cognitive biases in combination. We could see that status quo and
confirmation biases had led us to assume the world would always be what it
used to be. Beyond that, we neglected to heed the wisdom of portfolio theory
that you shouldn’t lay all your eggs in one basket. We not only laid them in
the same basket, but also within a very short period of time—the last billion
was committed before the construction period of the first billion had been
finalized. If we had stretched this whole €10 billion program out over a
longer period, say 10 or 15 years, we might still have lost maybe €1 billion or
€2 billion but not the amount we incurred later.

We also saw champion and sunflower biases, which are about hierarchical
patterns and vertical power distance. Depending on the way you organize
decision processes, when the boss speaks up first, the likelihood that
anybody who’s not the boss will speak up with a dissenting opinion is much
lower than if you, for example, have a conscious rule that the bigwigs in
the hierarchy are the ones to speak up last, and you listen to all the other
evidence before their opinion is offered.

31
 And we certainly overestimated our own abilities to deliver, due to a good
dose of action-oriented biases like overconfidence and excessive optimism.
Our industry, like many other capital-intensive ones, has had boom and
bust cycles in investments. We embarked on a huge investment program
with a whole generation of managers who hadn’t built a single power plant
in their professional lives; there were just a few people left who could really
remember how big investments were done. So we did something that the
industry, by and large, hadn’t been doing on a large scale for 20 years.

The Quarterly: On the sunflower bias, how far down in the organization do you
think that went? Were people having a hard time getting past their superiors’
views just on the executive level, or all the way down?

Bernhard Günther: Our investigation revealed that it went much farther

down, to almost all levels of our organizational hierarchy. For example,
there was a feeling within the rank and file who produced the investment
valuations for major decisions that certain scenarios were not desired—that
you exposed yourself to the risk of being branded an eternal naysayer, or
worse, when you pushed for more pessimistic scenarios. People knew
that there were no debiasing mechanisms upstairs, so they would have no
champion too if they were to suggest, for example, that if we looked at a
“brilliant” new investment opportunity from a different angle, it might not
look that brilliant anymore.

The Quarterly: So, what kind of countermeasures did you put in place to tackle
these cultural issues?

Bernhard Günther: We started a cultural-change program early on, with

the arrival of our new CEO, to address our need for a different management
mind-set in light of an increasingly uncertain future. A big component of that
was mindfulness—becoming aware of not only your own cognitive patterns,
but also the likely ones of the people you work with. We also sought to embed
this awareness in practical aspects of our process. For example, we’ve now
made it mandatory to list the debiasing techniques that were applied as part
of any major proposal that is put before us as a board.

It was equally important for us to start to create an atmosphere in which

people are comfortable with a certain degree of conflict, where there is an
obligation to dissent. This is not something I would say is part of the natural
DNA of many institutions, including ours. We’ve found that we have to
push it forward and safeguard it, because as soon as hierarchy prevails, it can
be easily discouraged.

32
 So, for example, when making big decisions, we now appoint a devil’s
advocate—someone who has no personal stake in the decision and is senior
enough in the hierarchy to be as independent as possible, usually a level
below the executive board. And nobody blames the devil’s advocate for
making the negative case because it’s not necessary for them to be
personally convinced; it’s about making the strongest case possible. People
see that constructive tension brings us further than universal consent.

The Quarterly: How did you roll all this out?

Bernhard Günther: There were two areas of focus. First, over a period of
two years, we sent the top 300 of our company’s management to a two-week
course, which we had self-assembled with external experts. The main
thrust of this program was self-awareness: being more open to dissent, more
open to a certain amount of controlled risk taking, more agile, as with rapid
prototyping, and so forth.

RAPID REFLECTIONS
FROM BERNHARD GÜNTHER

1 ADVICE
IN YOUR EXPERIENCE, WHAT PIECE OF COMMON LEADERSHIP
IS WRONG OR MISLEADING?
People development based on weaknesses—or gaps versus “ideal candidate”
profile—instead of building on strengths

2 WHICH HISTORICAL FIGURES DO YOU ADMIRE THE MOST?

Nelson Mandela and Martin Luther King Jr.

3 WHAT’S THE BEST BOOK YOU’VE READ IN THE PAST YEAR?

Freedom, by Jonathan Franzen (fiction)

You! The Positive Force in Change: Leveraging Insights from Neuroscience

and Positive Psychology, by Eileen Rogers and Nick van Dam (nonfiction)

4 WHAT SKILL DO YOU THINK IS MOST UNDERVALUED IN

LEADERS TODAY?
Listening

33
 Then we also launched a training program for managers and experts,
especially those involved in project work—for example, the financial
controllers that have to run the models for big investment decisions. This was
a combination of a training course, some desktop training you could do on
your own, and some distributed materials.

This program explicitly focused on debiasing. It started with these typical

examples where you can show everybody how easily we fall into those
cognitive traps, framing it not as a personal defect but as something that’s
just there. Secondly, it emphasized that debiasing can be done much more
easily within a group, because it’s a collective, conscious effort. And not some
kind of empty ritual either. We taught very specific things that people could
apply in their daily practices. For example, you can do a kind of premortem
analysis and ask your team, “Imagine we are five years into the future, and
this whole project we’re deciding on today has turned out to be a complete
disaster. What could have happened in the meantime? What could have gone
wrong?” This is something that we are now doing regularly on big projects,
especially when there are uncertain environmental factors—whether
macroeconomic, technological, ecological, or political.

The Quarterly: Could you tell us about an example or two where you made a
different decision as the result of debiasing practice, where it went the other way
from what you initially thought was the right answer?

Bernhard Günther: Two examples immediately come to my mind. The first

one came up in the middle of 2015, when it became obvious that our company
was in a strategic deadlock with the power-generation business—the cash
cow of the company for years but now with a broken business model. There
was a growing awareness among senior management that trying to cure
the crisis with yet another round of cost cutting might not be good enough,
that we needed to consider more radical strategic options. We established
a red team and a blue team to come up with different proposals, one staffed
internally and one with externals. We wanted an unbiased view from the
outside, from people who were not part of our company or industry; in this
case, we brought in external people with backgrounds in investment banking.

The internal team came up with the kind of solution that I think everybody
was initially leaning toward, which was more incremental. And the external
team came up with a more disruptive solution. But because it was consciously
pitched as an independent view, everybody on the board took their time
to seriously consider it with an open mind. It planted the seedling of the
strategy that we adopted to split the company into two parts, which now,

34
 a good year later, has successfully concluded with the IPO of Innogy. If we
hadn’t taken this approach, maybe months later or years later, somebody
would have come up with a similar idea, but it wouldn’t have happened that
fast, with that kind of momentum.

The second example is a recent potential investment project in renewable

energy that carried high reputational value for us, so there were emotional
issues attached to winning the project. We were bidding for a wind park that
was to be built, and the lowest bidder wins by offering the lowest electricity
price. We knew it would be a very competitive auction for that project, and we
had already decided in the run up to the decision making that we wanted to
have a devil’s advocate involved.

We had the project team make the case first in the board meeting. Then we
had the devil’s advocate put forward analysis of the risk–return trade-offs.
All of this was in written form, so everybody had to read it before the meeting.
This certainly helped our discussion a lot and made it much easier to have a
nonemotional debate around the critical issues. And we came out of it with a
different and I think better decision than we would have if we had just taken
the proposal of our internal project team at face value.

The Quarterly: Now that these decision-making changes have taken hold, how
do you see things running differently in the organization?

Bernhard Günther: Looking back at where we were three or four years ago,
I’d say that this practice of awareness and debiasing has now become almost
a part of our corporate decision-making DNA. But it’s something you have
to constantly force yourself to practice again and again, because everyone at
some point asks, “Do we really need to do it? Can’t we just decide?” It’s a very
time-intensive process, which should be utilized only for the most important
decisions of strategic relevance. About 30 percent of our board’s decisions
fall into this category—for example, major resource-allocation decisions—
and it’s similar elsewhere in the company.

Also, people’s general awareness of the complex set of issues around cognitive
biases has grown dramatically. Before this, things easily degenerated into
blaming exercises going both ways. The naysayers were critiquing the others
for wanting to push their pet projects. And the people promoting these
projects were saying that the naysayers were just narrow-minded financial
controllers who were destroying the company by eternally killing good
business ideas. But now there’s more mutual respect for these different
roles that are needed to ultimately come up with as good a decision outcome
as possible. It’s not just about debiasing; it’s given us a common language.

35
 It’s now routine for somebody to say in a meeting, “I think we need some
debiasing here.” And then everybody can agree to this without any need to
get emotional. When in doubt, we just go through the process.

The Quarterly: Do you have any recommendations for other senior leaders who
might be reading this interview?

Bernhard Günther: I think when you read about these issues, it can seem a
bit esoteric. You might say, “Well, maybe it’s just their problem, but not mine.”
I think everyone should just do it; just start with it even on a pilot basis. You
don’t have to start rolling it out across 1,000 people. You can start with your
own board, with a few test examples, and see if you think it helps you. But if
you do it, you have to do it right; you have to be serious about it. Looking back,
there were a few key success factors for us. For one, top management has to
set an example. That’s true of any kind of change, not just debiasing. If it’s not
modeled at the very top, it’s unlikely to happen further down the hierarchy.
Second, everyone has to be open to these ideas or it can be difficult to really
make progress. At first glance, many of the tools might seem trivial to some,
but we found them to have a very profound effect.

Bernhard Günther joined RWE in 1999 and served as the company’s chief financial officer
from 2013 until the 2016 spin-off and IPO of Innogy, where he is now CFO. This interview was
conducted by Sven Heiligtag, a partner in McKinsey’s Hamburg office, and Allen Webb,
McKinsey Quarterly’s editor in chief, who is based in the Seattle office.

Copyright © 2017 McKinsey & Company. All rights reserved.

36
GET YOUR WEEKLY DOSE OF SMART
Sign up for weekly email insights
from the Five Fifty

Subscribe

Or go to:
mckinsey.com/Quarterly/The-Five-Fifty

The management knowledge you need,

in quick briefings and select deeper dives.

Separate the signal from the noise—with the

McKinsey Quarterly Five Fifty.

Share this edition of the Five Fifty