Where to Start

If you have not already, start with A Closer Look.

Eighteen questions about what you can actually observe where you work, about seven minutes, no score and no grade. What it gives you is narrower and more useful than this chapter: the three or four chapters that match whatever is currently creaking, rather than a general answer to a general question. Working out which parts of a paper this size are yours should not be your job.

Then come back here, because knowing where to look and knowing what to do on Monday are different problems.

This is the one chapter in the paper allowed to sound like a checklist, and it should, because a reader who is convinced and doesn’t know what to do next was lost for an avoidable reason.

This week. Pick one capability that matters. Ask three people in different parts of the organization who owns it. Don’t announce that you’re doing it, and don’t correct the answers — just collect them. Whatever you find is the most accurate diagnostic in this paper — and asking is already the smallest version of doing something about it, since three people are now holding a question they weren’t holding on Monday.

This month. Time the next significant cross-team decision. Record when it became necessary, when it was made, and how much of the gap was deliberation. Bring both numbers to a leadership meeting without a recommendation attached. The number does the arguing.

This quarter. Take one thing you are currently funding as a project and describe it as a capability instead — what must we always be able to do, and is it better than last year? Notice how much of your reporting stops working when the question changes. That difficulty is the finding.

This year. Take one boundary — between two teams, two departments, two systems — and ask what it costs in handovers, waiting, and rebuilt context. Then ask whether the boundary is where it is for a reason that still holds, or because of a decision made in a room a long time ago by people who were solving a different problem.

And one that isn’t small, if the authority happens to be yours. Take something currently funded as a project with an end date, and fund the capability behind it for two periods instead. That is a single decision, it cannot be delegated downward, and no amount of good work further down substitutes for it — a team can be perfectly clear about who owns an outcome and still be dissolved on schedule before the outcome arrives. Funding What Doesn’t End is the long version.

The first four are deliberately small enough to do without permission. This one is deliberately not, and if you are reading the other four and concluding they are somebody else’s task, this is the one that was written for you.

None of these are transformations. That’s deliberate. The first four are small enough to do without permission, and each one produces a fact rather than an opinion — which, in an organization where the friction is invisible, is the only thing that reliably changes anyone’s mind.

What to do with what you find

Collecting the fact is the easy half. What happens next decides whether any of it was worth doing, and there are three ways it usually goes wrong.

Don’t fix it quietly. The temptation, having found that nobody owns something, is to sort it out yourself — because you can, and because raising it feels like making trouble. Do that and you have removed the evidence along with the symptom. The organization learns nothing, and the same gap reappears somewhere you aren’t.

Don’t present it as an indictment. Every number you collect has people attached to it. The decision that took eleven weeks was waited on by someone who could have unblocked it, and that person is likely in the room. Present the wait as a property of the system, because it is one — nobody designed the eleven weeks, and nobody chose them.

Don’t propose the reorganization. The instinct after finding unclear ownership is to redraw boundaries, and it is almost always premature. The first move is to make one existing owner unambiguous and see what stops being needed. Structural change is expensive, slow, and burns the credibility you will need for the second finding.

How to put a number in front of people

The numbers in this paper are small, personal, and easy to dismiss. Presented badly they sound like a complaint with arithmetic attached. A few things make the difference.

Bring the distribution, not the average. One decision that took fourteen weeks tells a room more than a mean of five, because everyone present can remember the fourteen.

Attach it to something already agreed. Nobody funds “reducing decision latency.” They do fund the initiative that is late because of it. The number is a diagnosis of something the organization already cares about, and should be introduced that way.

Say what it would take to be wrong. Volunteering the weakness of your own measurement is the fastest route to being believed. It is also honest: these are small samples, self-reported, and directionally useful rather than precise.

Ask for the second measurement, not for the change. The first number proves nothing on its own. Getting agreement to take it again in three months is a much smaller ask than getting agreement to act, and it commits the organization to noticing.

What to expect

Less than you want, later than you’d like.

The first measurement will be dismissed by someone as unrepresentative, and they will have a point. The second one will be dismissed too. Somewhere around the third, the conversation changes — not because the evidence became overwhelming, but because a number that keeps coming back stops being an argument and starts being a condition.

Expect the line to move both ways. A reading that gets worse after a boundary moved is the instrument working, and a quarter where nothing improves says more than one where everything does. What you are watching is not a number that only rises. It is whether the organization is still noticing.

That is the honest timescale, and it is worth saying plainly rather than promising a workshop that fixes it. Organizations change at the speed of the loop they can actually close.

What it costs you

No chapter in this part has said what any of this costs the person doing it, and a paper that asks people to work differently without pricing it reads as free, and therefore as unlikely.

It costs visibility.

The evidence that this is working is an absence. A meeting that stopped happening. A wait nobody experienced. An escalation that never arrived, a decision that did not need you, an incident that did not occur. Absences do not report themselves, do not appear on a scorecard, and cannot be described afterwards without sounding like a claim about a hypothetical.

So anyone doing this well will have a quieter quarter than a colleague who staged a visible rescue, and will be less legible in the room where those things get discussed. That is the actual trade, and it is worth knowing about in advance rather than discovering it in March.

I am not going to dress that up. It is the same pattern No Incident, No Story describes, applied to the person reading this — and the only honest mitigation is to be deliberate about the record: write down what you expected before you change anything, so that later you have a before and an after rather than a feeling.

The compensation, such as it is, is that this work is unusually durable. The forum you removed stays removed. The ownership you settled tends to stay settled long after anyone remembers it was ever unclear, which is both the reward and the reason nobody will thank you for it.

So start where you can already see something. What is hidden further in tends to become visible once the first thing moves, and rarely before.