When Good Decisions Go Wrong Before the Decision Is Even Made
A rigorous evidence base cannot rescue an inquiry whose question, boundaries, assumptions or perspectives were narrowed too early.
A rigorous evidence base cannot rescue an inquiry whose question, boundaries, assumptions or perspectives were narrowed too early.
A decision can be analytically rigorous and still have gone wrong before the analysis begins.
Consider a fairly ordinary organisational problem.
A service is underperforming. Senior leaders ask:
“How do we improve compliance with the existing process?”
A team is assembled. Data is gathered. Experts are consulted. Options are assessed. Risks are recorded. Recommendations are written. The eventual decision may be thoughtful, evidence-based and entirely rational in relation to the question that was asked.
But look again at the question.
It already contains several decisions.
It assumes that poor compliance is the problem. It assumes the existing process should remain. It directs attention towards the behaviour of people using the process rather than the design of the process itself.
Compare it with:
“What is preventing this service from achieving the outcome we want?”
The two questions could lead to very different inquiries.
Neither is automatically correct. That is the point.
The wording of the question has begun shaping the answer before anyone has examined the evidence.
The first decision may be the question
We tend to think of decision-making as something that happens towards the end of a process.
First we define the problem. Then we gather evidence, conduct analysis, consult people, develop options and finally make a decision.
In reality, consequential choices occur much earlier.
Someone decides what the problem is.
Someone decides what is inside and outside the inquiry.
Someone decides whose expertise is relevant.
Someone decides which assumptions are sufficiently safe to proceed with.
Someone decides which possibilities deserve investigation and which appear too implausible or peripheral to pursue.
These choices are often reasonable. Some are unavoidable. No inquiry can consider everything.
The difficulty is that they do not always look like decisions.
Once they disappear into the foundations of an inquiry, everything built on top of them can appear objective.
This is one reason framing matters. Amos Tversky and Daniel Kahneman demonstrated more than four decades ago that changing the way an otherwise equivalent decision problem is presented can alter people's preferences.
In complex organisational decisions, the issue can be subtler still. We are rarely choosing between two neatly defined alternatives. We are also deciding which alternatives should exist in the first place.
What never entered the room?
Suppose ten capable people spend a day examining a difficult strategic question.
They challenge one another. They use good data. They document their conclusions. Nobody dominates the conversation.
There can still be a serious weakness.
What if the perspective that would have challenged the central premise was never represented?
This is not simply about stakeholder representation. It can be about analytical perspective.
Perhaps a technically sound proposal was never seriously examined through an operational lens.
Perhaps an economically compelling option was not tested from the perspective of the people expected to use it.
Perhaps a policy question was considered by policy specialists, analysts and senior leaders, but nobody examined the behavioural assumptions on which it depended.
The resulting answer might accurately reflect everything considered in the room.
The difficulty lies in what never entered it.
That creates an important distinction between a balanced discussion among the people present and a sufficiently balanced inquiry into the problem itself.
They are not the same thing.
Assumptions are useful. Invisible assumptions are dangerous.
Every serious inquiry relies on assumptions.
Time is limited. Evidence is incomplete. Future behaviour cannot be known with certainty. Some things have to be treated as provisionally true so that work can proceed.
The aim should not be to eliminate assumptions.
It should be to recognise them.
There is an important difference between saying:
“We expect demand to remain broadly stable over the next three years.”
and allowing stable demand to become an invisible property of the analysis.
In the first case, the assumption can be challenged, tested and revisited.
In the second, it quietly shapes everything that follows.
Established analytical practice already recognises this problem. The UK Government's AQuA Book recommends recording assumptions explicitly and treats scoping as a distinct part of analytical work.
That matters because visible assumptions can become objects of inquiry.
Invisible assumptions become architecture.
Excellent evidence can answer the wrong question
Evidence matters enormously.
But evidence enters a structure that already exists.
If the original question is too narrow, better evidence may simply provide a more confident answer to the narrow question.
If an important option was excluded at the outset, detailed analysis of the remaining options will not bring it back.
If everyone accepts the same unexamined assumption, adding another expert who shares it may strengthen confidence without increasing understanding.
The UK Government's Green Book similarly encourages practitioners to generate a sufficiently broad range of options and to record which options were considered and why some were not progressed.
That raises a broader question.
How much of the reasoning that shapes an inquiry should remain visible throughout it?
Not just the final evidence.
The question.
The scope.
The assumptions.
The alternative interpretations.
The perspectives applied and those that were absent.
The uncertainties that could not be resolved.
And the reasoning that connects all of these to the eventual answer.
A defensible answer is more than a confident one
I have become increasingly interested in this front end of decision-making because difficult questions rarely fail through a simple shortage of intelligence.
Very capable people can produce poor outcomes.
Very good evidence can coexist with poor framing.
A room full of experts can still share the same blind spot.
And a conclusion can look remarkably robust if the uncertainties and disagreements that preceded it have disappeared by the time it reaches the decision-maker.
This does not mean every decision requires months of analysis or a vast consultation exercise. Structure should be proportionate to consequence and uncertainty.
Sometimes the right answer really is obvious.
But when an issue is contested, consequential or uncertain, perhaps we should apply as much discipline to constructing the inquiry as we do to analysing the evidence that follows.
That is one of the questions that led me to start developing INQUIRIM.
I am exploring whether inquiry itself can be made more structured. Not as a machine for producing a supposedly “right” answer, but as a way of reaching a good, considered, impartial and evidenced one while retaining enough of the underlying reasoning for somebody else to understand how we got there.
There is a great deal still to test.
For now, I am interested in the problem that comes before any proposed solution.
When you look back at difficult decisions you have seen, where did the real constraint first appear?
Was it in the evidence?
Or had the direction already been shaped by the question, the scope, an assumption, or a perspective that nobody realised was missing?
I would be particularly interested in public examples or professional experiences that can be discussed without revealing anything sensitive or confidential.