The decision you cannot outsource to data

You have everything the data can give you. The call is still entirely yours.

The decision you cannot outsource to data

This is the toughest decision you have faced in a while. You really do not enjoy these, but you know it has to be made, and it has to be made by you.

You are sat in your office with the analysis in front of you. It is excellent, everything you asked for and more. The performance history, the assessment scores, the calibration against peers, the track record across three roles, the pattern of how this person has handled pressure before. The team that prepared it has done a good job, and you have read all of it, twice.

The decision is whether to back this person for the role that is coming. A bigger seat, more exposure, a step up in consequence. And sitting with the complete picture, every relevant number present, you realise that the thing you are actually being asked to decide is not anywhere in it.

Whether this person is right for what is coming. Whether you trust them with it. Whether they will grow into the seat or be found out by it. And where there is stretch, as there always is in a decision worth agonising over, the real question is not whether they can develop. It is whether they will develop fast enough to make the impact the role will demand of them, on the timescale it will demand it. None of that is in the data. It was never going to be.

You could ask for more. Another assessment, another calibration, another round of input. And some part of you knows that none of it would make the decision, because the decision is not the kind of thing the data can make. You have everything the data can give you. The call is still entirely yours.

That moment, the complete analysis and the decision that is not in it, is the subject of this article.

What the data has genuinely done

Begin by giving the data its due, because this is not an argument against it.

The analysis in front of you is not noise, and it is not a comfort blanket. It has done real work. It has told you what happened, with a precision that would have been impossible a generation ago. It has surfaced patterns you would have missed, corrected for biases you would have carried, ruled out options that felt right and were wrong. A leader who ignored it would be foolish, and the leaders worth doing the analysis for do not ignore it. They commission it, read it properly, and take it seriously.

And the better it gets, the more this matters. As the tools for quantifying, modelling, and predicting become more powerful, as more of what used to be guesswork becomes measurable, the data settles more of the question than it ever could before. More of the ground is mapped. More of the decision is, genuinely, made for you by the analysis, and made better than you would have made it alone.

This is the lens worth carrying, and it is not what you might expect. The residue beyond the data. The part of the decision that is left after the analysis has given everything it has, the part no quantity of further data will reach. The striking thing, the thing leaders feel but may not be conscious of, is that this residue does not shrink as the analysis improves. It sharpens. The more the data settles, the more clearly you can see the exact shape of the part it was never going to settle, standing out against everything it has resolved around it.

The data can tell you everything except the one thing you are being paid to decide.

I have sat with this more times than I’d like, and from the chief executive's chair as well as alongside it. A complete analysis in front of me, prepared well, every number I had asked for, and the knowledge that the decision the moment actually required was not in any of it, and that no further work would put it there. The temptation, always, was to ask for one more piece of analysis. The discipline was to recognise when the analysis was already complete, and the thing left undecided is yours to decide.

Why the residue cannot be closed

It would be reasonable to assume the residue is just the part the data has not reached yet, that better tools, more inputs, a richer model would eventually close it. Sometimes that is true, and where it is true, get the better analysis. But the residue this article is about is not a gap in the data. It is a different kind of thing altogether.

The decision about the person is not unquantified because no one has quantified it. It is unquantifiable because it is a judgement about a future that does not exist yet, made about a human being who is not fixed, in a situation that has no exact precedent. Whether someone will grow into a seat, and whether they will grow fast enough, is not a fact waiting to be measured. It is a judgement about how a particular person will meet a particular set of pressures that have not arrived, at a pace the situation will set rather than one they can choose, and the data, however complete, is all drawn from a past that is not the future they are being asked to step into.

The same is true of the genuinely strategic call, the commitment, the bet, the decision to enter a market or leave one, to acquire, to change course while the existing course is still working. The models can tell you everything about the world as it has been and a great deal about the world as it is. What they cannot tell you is what to do, because what to do is a judgement that has to weigh things the model holds equal, has to value a future against a present, has to decide what kind of organisation you intend to be and what you are willing to risk to become it. That is not in the data because it is not the kind of thing data contains.

This is why the residue does not yield to more analysis. You are not waiting for a number that has not arrived. You are holding a decision that is structurally not the data's to make, and mistaking it for one that is keeps you reaching for analysis long after the analysis is done.

The leader who has learned where the data stops

There is a kind of decision-maker who has learned, the hard way and at the highest stakes, exactly where the data stops, because the cost of confusing the two has been paid in full at least once.

Think of those who carry decisions of real consequence for many people, the leader of a large institution, the chief executive whose call will land on thousands of employees, the leader whose decision shapes communities or markets well beyond the walls of the organisation. They have, very often, the best analysis money and expertise can produce. Whole functions exist to give them the numbers. And the ones who have done the job for any length of time know, in their bodies, that the analysis takes them to the edge of the decision and no further, that the part that matters most, the weighing of consequence, the judgement about people and futures, the call about what is right rather than what is merely optimal, is theirs and cannot be handed to the model.

What they have learned is not to trust the data less. They trust it completely, for what it is. What they have learned is to recognise the precise point at which the data has given everything it has, and to stop looking to it for the thing it cannot give, because the habit of reaching for one more analysis is, at that point, a way of not deciding.

You may carry consequence at exactly this scale, or at a scale that feels just as heavy in your own chair, where the decision lands on people you know by name. The weight is not reserved for the largest institutions. Wherever you sit, the boundary is the same. The data takes you to the edge of the decision. The decision is on the other side of the edge, and it is yours.

What it costs to keep reaching

When a leader will not accept that the analysis is complete and the call is theirs, the cost often shows up in these four places.

It shows up first as decisions that arrive too late, and as the actions of others filling the space you left open. The window for a good call is rarely open indefinitely. The leader who keeps commissioning another analysis to avoid the unquantifiable part often finds the decision has been made for them, by time, by a competitor who moved, by the key person who grew tired of waiting and made their own decision to leave, by a board that acted, by a market that decided. The decision still gets made. It is simply made by others, in a worse form than the one you could have chosen, and from a position you no longer control.

It shows up second as analysis used as an alibi. A decision made strictly from the data can always be defended, because the numbers said so. This is a comfort, and it is corrosive, because it lets a leader avoid the part of the decision they are actually paid for, the judgement, and hide the avoidance behind a methodology. The most over-analysed decisions are often the ones the leader most wanted to avoid owning.

It shows up third as a team that learns the data decides. When the leader will only act on what the numbers settle, the people around them learn to bring numbers and withhold judgement, and the organisation slowly loses its capacity for the kind of thinking the data cannot do. You train the people you lead in what you reward, and a leader who rewards only the quantifiable gets an organisation that has forgotten how to weigh anything else.

It shows up fourth, and most quietly, as a leader who is gradually deskilled in their own core work. Judgement is a capability, and capabilities decline when they are not used. A leader who routes every decision through the data, and treats the residue as a problem to be analysed away rather than a judgement to be made, finds over time that the muscle for the unquantifiable call has weakened, exactly when the stakes are highest and the data is least able to help.

A practice in three directions

When you are sitting with a complete analysis and a decision that is not in it, the work moves in three directions.

Honour what the data has settled. Start by being precise about what the analysis has genuinely done, because the residue only becomes clear once you have given the data its full due. What has it ruled out. What has it established that you would otherwise have got wrong. Where has it corrected your first reading, and rightly. Take all of it seriously. The leaders who carry the unquantifiable call well are not the ones who trust data least; they are the ones who have used it most wisely, so that what remains is genuinely only the part it cannot reach.

Name what the data cannot reach. Then say plainly, to yourself, what the decision actually turns on that is not in the analysis. The judgement about a person's future, and the pace of it. The weighing of one value against another. The call about what kind of organisation you mean to be. Naming it matters, because the unnamed residue is what keeps you reaching for more analysis. Once you have said clearly "this part is not in the data and will not be," you stop waiting for the data that is not coming.

Carry what is left, as judgement, not as a gap. What remains is not a deficiency in your information. It is the decision. It is the thing you are there to do, the reason the seat exists and the reason it is yours and not the model's. Carry it as that, a judgement to be made with everything you are, your experience, your reading of people, your sense of consequence, rather than as an uncomfortable gap you wish the data had filled. The residue is not what is missing from the decision. It is the decision.

A leadership reality

The data can tell you everything except the one thing you are being paid to decide.

This is not a reason to value the analysis less. The better your data, the better your decisions, right up to the edge of the part it cannot make, and that edge is further out than it has ever been. The discipline is not to distrust the numbers. It is to recognise the precise point at which they have given everything they have, and to meet the decision that is left without pretending it is the data's to make.

Your analysis is not the constraint. It is excellent, and it is getting better. The constraint is the quality of judgement you bring to the part of the decision the data was never going to reach, and that judgement is not a fixed trait. It can be developed, and in a world where more and more of the quantifiable is handled for you, it is fast becoming recognised as the part of leadership that matters most.

One question worth carrying

Where am I reaching for more data to make a decision that more data will never make for me?

The question is worth asking of the decisions currently on your desk, the ones that have been waiting a little too long, the ones where you keep asking for another view. Most leaders find at least one where the analysis was complete some time ago, and what has been missing is not information but the willingness to make the call that was always going to be theirs. Finding it is not a failure of rigour. It is the recognition that the rigour has done its work, and the judgement is now yours to bring.

It is a wise leadership move to know the difference between the decision that needs more analysis and the one that needs you.

May you always find wise judgement when certainty is unfindable.