Decision Making Under Uncertainty

Decision
Making
Under
Uncertainty
Taking the bias out of
decision making
– Unilever’s story
Andrew Evans
Think Clearly – Act Decisively – Feel Confident
What we will cover
Introduction to Unilever
and the role of
Decision Analysis
Why are we
biased?
Some steps to take
the bias out of
decision making
Unilever is a large, culturally diverse and
complex organisation
Some of our top brands:
Global presence
2010 Turnover ~€44 billion
Americas €14 bn
• 167,000 employees
• Products sold in more than
180 countries
• Large degree of cultural
diversity
Europe €12 bn
Asia / Africa €18 bn
“2 billion consumers
worldwide use a Unilever
product on any day”
Decision Analysis is being actively applied
across Unilever, in areas such as…
… Innovation, Mergers & Acquisitions, Supply Chain, Safety, Regulatory
and other complex one-off decisions
All these applications have common features: there are multiple
compelling alternatives; there are significant contradictions /
disagreements on how to proceed; and stakes are high
The Unilever DMUU programme is embedding
Decision Analysis throughout the organisation
Framing & Structuring
•
•
•
•
Clarifying the problem
Creating alternatives
Structuring information
Improving communication
Analysis
•
•
•
•
Quantifying uncertainties
Testing assumptions
Evaluating alternatives
Identifying sources of value
Insight and commitment to action
Creating
Creating aa culture
culture where
where we
we have
have the
the debate,
debate, make
make aa
decision
decision and
and move
move on
on as
as aa team,
team, even
even when
when we
we have
have to
to
make
make tough
tough choices
choices with
with limited
limited information.
information.
Can you see what it is yet?
Our ability to recognise and learn patterns has
been an important part of our success story!
Our individual personalities give us unique
perspectives!
Our offices, as seen by
someone with a strong
detail focus…
…and by someone with a
more abstract focus!
These perspectives make us susceptible to different
biases
There is a wide range of decision making biases…
“I’m an above average driver”
“Of course my product launch
will be a success”
Confidence is important, but overconfidence can be dangerous
…and further decision-making biases are
constantly being identified
Sunk cost
Status Quo
Selective
perception
Framing
Despite its tendency to bias, a human brain is
excellent at:
Recognising and remembering
patterns.
Detecting differences in views.
NPV (W$m)
Comparison of NPV distribution for four strategies
5.1
-3.8
8.0
7.7
Drawing insights from analysis.
We need a way of exploiting the advantages our brains
give us, whilst avoiding the downsides!
Suggestion #1 – Understand the decision, and
understand yourself!
Increasing Complexity
Low Complexity
When operating in this area:
1. Consider your own biases
2. Consider ‘decision fitness’:
• Overtired?
• Stressed?
• Not well?
3. Use the Decision Quality
‘Quick test’
High Complexity
At high complexity levels,
use an appropriate process
to drive out biases
The level of decision complexity will determine the
best approach
The Decision Quality ‘Quick test’ helps in lowcomplexity situations
Appropriate frame – what
question do we want to
answer?
What is the full range
of alternatives to
consider?
What information do we
need and where can we
get it from?
Logical reasoning – have
we reliably assessed each
alternative?
Criteria & trade-offs –
how will we decide
between alternatives?
Who is the decision
maker and will they
commit to action?
Suggestion #2: Use an effective process for
complex decisions…
Decision Makers
Review 1
Review of framing
and structuring.
IDENTIFY
PROBLEM
FRAME
Review 2
Review of key data
and confirm frame is
still valid.
STRUCTURE
Review 3
Discussion of
analysis, agree
decision and actions.
ANALYSE
Evaluation Team
RECOMMEND
…and choose tools that will compensate for
team member’s biases
Selective perception
Run a brainstorming session to force new concerns to be considered
Create a range of strategies to ensure a full range of alternatives is covered
Status Quo
Build several strategies to ensure other options are considered
Fully explore each strategy – get others to comment on the benefits
and disadvantages of each one
Framing
Explore multiple frames at the start of the process and document the
agreed decision frame
Re-examine the frame at stages throughout the process
Sunk cost
Use communication aids such as decision trees to identify past
and future costs
Suggestion #3 – Focus your presentations
“Modified Original strategy had an EV of W$ -3.8m,
whereas Step by Step strategy was W$8.9m better
with an EV of W$ 5.1m. National Factory now
showed a 56% improvement over Step by Step but
only a 4% improvement over National Factory Later
strategy which had an EV of W$ 7.7m.”
NPV (W$m)
Comparison of NPV distribution for four strategies
“a picture is worth
a thousand
words”
5.1
-3.8
8.0
7.7
Suggestion #3 – Focus your presentations
48% chance of negative NPV
Your senior decision
maker may not need all
of the technical content!
Mean = W$ 7.7m
!
NPV (W$m )
Our goal is better decisions – despite the biases!
Our brains are excellent at recognising patterns and drawing
insights. But this strength makes us all prone to biases. If we
are aware of them, we can limit their impact on decision quality
In low complexity situations we can use a checklist to
avoid some common biases
Decision Makers
In more complex decisions a more thorough decision
process, with careful choice of tools is appropriate
Review 1
Review of framing
and structuring.
IDENTIFY
PROBLEM
FRAME
Review 2
Review of key data
and confirm frame is
still valid.
STRUCTURE
Review 3
Discussion of
analysis, agree
decision and actions.
ANALYSE
Evaluation Team
We can tailor the insights from our analysis to
the needs of our decision makers
RECOMMEND