Management foundations: four standing obligations and what you cannot delegate

Ask a manager what they did last week and you will usually get a list of meetings. Ask what changed because of them and the answer takes longer.

The output of management is not an artifact. It is other people doing work that would otherwise have stalled, been done twice, or been done well in the wrong direction.

That makes the job hard to see and easy to fake. A manager who is busy looks the same as a manager who is effective, and the difference only shows up in how the team performs when that person is unavailable.

What to take away

The four standing obligations

Strip out the activity and the job reduces to four things you owe the people you manage. Everything on your calendar either serves one of them or is overhead.

Direction. They know what they are trying to achieve, why it matters, and what it is more important than. Not the company mission: the specific thing that this quarter is for, and what they should drop when it conflicts.

Capability. They can actually do it, and if they cannot yet, someone has taken responsibility for closing that gap on a stated timescale. This includes access, permissions, tools, information, and the authority to decide things they should be deciding.

Conditions. The environment does not fight them. Work arrives with what it needs, handoffs are defined, priorities do not reverse weekly, and the obstacles they cannot remove are removed by someone who can.

Consequences. Good work is recognized specifically, poor work is addressed early, and both happen predictably enough that people can calibrate. This is the obligation most often deferred, and deferring it is what makes the other three unenforceable.

When a team goes wrong, ask which of the four is missing; it is almost never all four, and the fix for one is not the fix for another.

Unclear direction looks like busy people producing the wrong thing. Missing capability looks like slow, careful work with lots of checking. Bad conditions look like competent people apologizing for delays that were not theirs. Absent consequences look like a team that has stopped noticing the difference between good and adequate.

The unit of the work is a decision someone else could not make

Not every question that reaches you is yours. Most are not. The ones that genuinely belong to you share three properties: they cross a boundary the asker cannot cross, they commit a resource the asker does not control, or they set a precedent that will bind other people.

Everything else is a question you should be answering once, structurally, rather than repeatedly, individually.

The test is simple and uncomfortable. When the same question comes a third time from different people, you have a missing decision, not a busy inbox.

Write the answer as a rule, say what it does not cover, and name who may grant an exception. Five minutes each time felt cheap, but the total is most of a working week, and the team cannot move without you.

The reverse error is worse. Pulling decisions upward that the team could make removes their ability to calibrate, and it teaches them to bring you everything, which then confirms your belief that they cannot decide.

Context is the thing you actually ship

Most of what a manager knows is never written down. Why the last approach was abandoned. What a particular stakeholder cares about. What the deadline is really driven by. Which part of the spec is negotiable. What happened the last time someone tried this.

People cannot make good decisions without that, and they will not ask for it because they do not know it exists. So it has to be pushed rather than pulled.

The practical version is a habit: explain the reason with the request. Not as courtesy, but because the reason lets someone adapt when the situation turns out differently than you both expected, which it will.

A person who knows only the instruction will follow it into a wall. A person who knows the constraint behind it will tell you the instruction no longer makes sense.

This is also the cheapest form of development available. Sharing the reasoning of a decision, including the parts you were unsure about, teaches judgment faster than any amount of feedback on outputs.

Attention is your only budget

Time is not the scarce resource. You can always add hours and get less done. What runs out is the number of things you can hold in mind well enough to notice when one goes wrong.

That number is small, and it does not grow with seniority. It sets a real limit on how many people you can manage, how many concurrent efforts you can sponsor, and how many problems you can personally carry.

Two consequences worth acting on.

First, delegation is not primarily about workload. It is about which problems occupy your attention, and a task you have delegated but are still worried about has not been delegated. Either give it away properly, with the authority and the acceptance condition attached, or keep it and admit that it is yours.

Second, when you take something on personally, something else silently loses your attention, and you will not choose which. So do the accounting explicitly: if you are stepping into a problem for a fortnight, name what you are stepping back from for that fortnight and tell the person affected.

Standards, preferences, and the difference that ruins teams

A standard is a line you will enforce every time, including when it is inconvenient, including for the person you like, including in the week the deadline is tight.

A preference is how you would do it.

Both are legitimate, but damage comes from presenting a preference as a standard or enforcing a standard inconsistently. Both make people unable to predict which rules are real.

Then they either check everything with you, returning every decision to your desk, or they guess, and half those guesses are wrong.

Practical discipline: keep the list of standards short enough that you will actually hold every one of them, state them as outcomes rather than methods, and when you are expressing a preference, label it as one. "I would have done it differently and it is fine" is a complete and useful sentence.

An exception granted without a stated reason converts a standard into a preference for everyone who hears about it. Exceptions are sometimes right. Silent ones never are.

What you cannot hand to anyone else

Almost everything is delegable. A short list is not, and the items on it are the ones most often avoided.

The middle two are where most managers quietly fail. A postponed conversation does not stay the same size: it accumulates, and by the time it happens it is about a pattern rather than an incident. That talk is harder and less fair.

The person on the other side almost always says they wish they had been told sooner.

Managing work you could not do yourself

At some point you manage people whose craft you do not have. The instinct is either to fake competence or to withdraw entirely, and both are worse than the honest position.

You can still ask what would have to be true for this to be the right approach, what the person is least sure about, and what they need to find out sooner.

Ask what they would do with twice the time, or half of it. None of that requires knowing the domain, and all of it surfaces the risk.

Stop evaluating the method; evaluate the outcome, the reasoning, and whether risks were named up front.

If you cannot judge technical quality, arrange for someone who can, and say plainly you are doing that. Pretending to assess what you cannot destroys credibility with the exact people whose trust you need.

How to tell whether any of this is working

Team metrics tell you about the work. These tell you about the management.

None of the five is a number, and that is deliberate. A management signal turned into a scored figure attached to a person stops describing the work and starts describing what people believe is being watched, which is the effect named in Goodhart's law.

For the wider set of operating practices these sit inside, the Small Business Administration's guide to managing a business is a reasonable starting point for the formal obligations that come with employing people.

What to hold on to

Management is mostly preparation. The decisions made once and written down, the context handed over before it is needed, the standard stated and then actually held, the difficult sentence said early. Do those and the daily work gets quieter, because most of what fills a manager's week is the cost of not having done them.

Common questions

How many people can one person manage well?

Fewer than most structures assume, and the number depends far more on the variety of the work than the headcount. Ten people doing near-identical work is easier than four people doing four different things. If you cannot say what each person is working on and what is at risk in it, you are past your limit regardless of the number.

Should a manager still do the work?

Some, deliberately chosen, for a reason you can state: staying close enough to judge quality, covering a genuine gap, or keeping a skill you will need. The failure is drifting back into it because it is familiar and produces visible output. The tell is that your own work is the part of your week that never slips.

What if I inherit a team with none of the four obligations in place?

Fix consequences last and direction first. Direction is the cheapest to establish and makes everything else legible, and enforcing consequences before people know what they are being held to is unfair and will be read as arbitrary.

How do I manage someone who is better at the job than I am?

By doing the parts they cannot do for themselves: removing obstacles, getting them the decisions and access they need, protecting their attention, and telling them honestly how their work is landing with everyone outside the team. Competent people rarely want to be taught. They want the friction gone.