The comparison that never ends
Comparative analysis helps compare assumptions. It does not remove uncertainty.
The document has been open for six weeks.
It started as a rough list of pros and cons. It has since become something more structured: criteria weighted by importance, options scored against each, a summary column at the bottom pointing in one direction. Two people have been consulted. A mentor has given a view. The numbers are clearer than they were.
The decision is not.
Each time the comparison feels close to finished, something shifts. A new variable surfaces. The weighting feels slightly off. One of the options has changed in a minor way that seems to warrant revisiting the whole thing. The document is more thorough than it was last week. The position of the person holding it is exactly the same.
This is not a failure of rigour. The analysis is genuine. The effort is real.
What is harder to see from inside it is that the comparison has stopped being a tool for reaching a decision and started being a substitute for it. The two feel almost identical. Both involve sitting with the problem. Both produce a sense of movement. The difference is that one is moving toward something and the other is rotating in place, carefully, in a way that is easy to mistake for progress.
Comparative analysis works well on a particular kind of problem. Choosing between mortgage products. Weighing up two software options. Evaluating suppliers against a clear set of requirements. These are bounded problems with relatively stable variables. The world does not change significantly while the comparison is running. The person making the decision does not change either.
Career decisions at any meaningful scale do not work this way and the comparison table tends not to flag this.
What the spreadsheet cannot hold
When someone is weighing whether to leave a stable role for something uncertain, they are not comparing two fixed states. They are comparing two futures that do not yet exist, from a position that will not exist either once the choice is made. The role being considered will be experienced by someone who has already committed to it. That is a different person from the one running the analysis from a safe distance, with both options still technically available.
What the spreadsheet cannot model is who the decision will make them.
A move into a more visible role, with higher stakes and less structural cover, does not just change the work. It changes who sees it and on what terms. It changes where failure lands when it occurs. Some people become sharper under that kind of exposure. Others find that what they wanted, the recognition, the room to operate, the sense of consequence, arrives alongside a discomfort they had not anticipated. The comparison described the upside. It did not describe what the role would feel like from inside it, once the excitement had settled into the ordinary rhythm of the week.
This is not a gap the analysis could have filled with more effort. It is a category of information that is only available from inside the experience.
Part of what makes the pattern durable is that the comparison table overweights what can be compared.
Salary is measurable. The quality of the environment being walked into is not. Scope is legible. The political texture of a new organisation, and how quickly it becomes visible, is not. Start date is easy to put in a column. Whether the people already doing this job seem like people who have been stretched or people who have been gradually worn down is harder to score.
So the measurable variables get louder. The harder observations get noted vaguely, or left out, because there is no clean way to include them. The table becomes more complete and less representative at the same time.
A person comparing two senior roles may carefully analyse compensation, reporting lines and growth trajectory. All of that matters. But the question that tends to shape the experience most is one that rarely appears in any column: what does this environment do to the people inside it over time? The answer is usually visible before the decision is made, in the conversations, in what does not get said, in how people who already hold the role talk about it. It requires a different kind of attention than comparison.
The cost of remaining uncommitted
There is a version of prolonged analysis that looks like conscientiousness and is sometimes mistaken for it, including by the person doing it.
The professional who has been sitting with a decision for four months has, in their own account, been thinking carefully. The effort is not fabricated. What is also true is that four months have passed in a position that requires no commitment and carries no cost. The analysis generates the feeling of movement while the actual decision stays where it was.
What tends not to be examined is what that interval costs.
Opportunities are not static. The role being considered may shift. The people attached to it move. The version of the self that would have developed by committing three months ago is not the same as the version produced by deciding now, after the energy of the original moment has dispersed.
The analysis never includes the cost of the analysis. That cost is real. It does not appear in any of the columns.
The more useful question, once the obvious variables have been mapped, is not which option scores higher.
It is what the comparison is protecting against.
In a number of cases, prolonged analysis is doing genuine work, holding open a real complexity that has not yet resolved. In others, it is providing a structurally respectable reason not to arrive at the moment where a choice would have to be owned. The person who has been refining a comparison for weeks while consistently finding reasons why the picture is not yet complete may be avoiding something that would not appear in any weighted criteria model: the possibility of having wanted something clearly enough to choose it, and then being the identifiable person who got it wrong.
Analysis used this way does not produce a decision. It produces increasingly detailed documentation of a decision that has not been made.
Different questions
What shifts the situation is rarely more analysis. It is a different set of questions, asked of the same material.
The first shift is from asking which option is better to asking which uncertainty is worth owning. These are not the same question. The first asks the comparison to produce a verdict. The second asks the person to locate themselves honestly inside the choice. It is less comfortable. It tends to move faster.
The second shift is from asking what could go wrong to asking which failure is liveable. Anticipated regret is a more reliable signal than probability estimates. People are generally better at knowing what they could live with having tried and lost than at predicting which outcome will materialise. The question is not which option is safer. It is which failure mode leaves the person more intact.
The third shift is from asking am I ready to asking what information is only available from inside the commitment. For identity-level decisions, a threshold of uncertainty is not a defect in the process. It is a permanent feature of it. Some things cannot be known before the decision is made. The comparison keeps running partly because it is searching for information that is structurally unavailable until after the choice.
The fourth shift is from modelling the future to briefly inhabiting it. A conversation with someone already doing the role, attended to carefully, is more predictively useful than a weighted criteria model. Not because it contains more information, but because it activates something the abstracted comparison cannot: a felt sense of what the experience is actually like, which the body registers before the analysis can process it. That signal is not decisive. It belongs in the picture.
There is also a distinction the comparison table tends to flatten: between decisions that can be revisited and decisions that cannot.
A significant amount of time goes into treating choices that could be tested in months as though they were permanent, because uncertainty is uncomfortable and discomfort is easy to mistake for consequence. If a path can be meaningfully revised within a year, it may not need another month of modelling. It may need the kind of information that only becomes available from inside it. Committing in order to find out is a different act from committing forever, and treating them as the same is one of the more reliable ways people stay exactly where they are.
The point of the comparison
Two people, same decision, six weeks apart.
The first has built a thorough comparison. They know which option scores higher. They have consulted the right people. They are waiting for one more piece of information that might, finally, make the picture feel complete.
The second stopped the comparison at the point where additional analysis was no longer changing the shape of the decision, only adding detail to a conclusion that was already forming. They asked which uncertainty they were prepared to carry, noticed which failure they could live with more honestly, and made the call. The information they were missing turned out to be available shortly afterwards, from inside the experience rather than before it.
The difference between them is not the quality of the analysis. It is what each person understood the analysis to be for.
None of this means analysis is the wrong place to start. It is often the right place to start. The problem is when it becomes the only move available, and the questions it cannot answer go unasked.
Most career decisions of any real weight contain a category of uncertainty that will not resolve before the decision is made. The comparison can map the known variables. It cannot close the gap between a future imagined from outside and a future lived from within.
At some point the document is as complete as it is going to be. The future is still unclear. The person still has to be the one who chooses, and who can be identified as having chosen, if it turns out to have been wrong.
What is required at that point is not a better spreadsheet. It is the willingness to choose under conditions that have not fully resolved, and to accept that some of what needed to be known was only ever going to be available afterwards.
The comparison table is preparation for that moment.
It is not a way around it.
Research notes
On affective forecasting: Daniel Gilbert’s research on how people predict their future emotional states shows consistent errors in both directions. Positive outcomes feel less sustaining than anticipated; negative outcomes feel less damaging. This structural limitation means that comparison, built on imagined future states, is working with systematically distorted material.
On the paradox of choice: Barry Schwartz’s work established that beyond a threshold, additional options and additional information increase anxiety and anticipated regret rather than improving decisions. Prolonged comparison raises the standard a decision must clear, making commitment harder without making the choice better.
On anticipating regret: Kahneman and colleagues found that people are more accurate at predicting what they will regret not doing than at estimating outcome probabilities. Regret minimisation, asking which failure is more liveable, tends to cut through the symmetry that extended comparison produces.
On identity-level decisions: Herminia Ibarra’s research on career transitions found that people who navigated major professional changes successfully did so through small experiments and provisional commitments rather than through planning. The self is partly constructed through action, not only revealed by it. Comparison assumes a stable self evaluating fixed options. The assumption does not hold for decisions that change who the person is.
On structural uncertainty: Acceptance and Commitment Therapy research distinguishes between uncertainty that resolves with more information and uncertainty that belongs to the structure of the situation. The relevant clinical finding for professional decisions is that treating structural uncertainty as a problem to be solved before acting tends to intensify avoidance rather than reduce it. The more productive orientation is to identify what matters enough to choose, even without knowing how it will turn out.
Human & Machine studies how judgement fails under complexity. This piece is part of that work.
Does this pattern feels familiar? You can explore it further through the Decision Audit.