If everyone manages themselves, what is left for the manager

A consulting engineering firm, 130 people, a mid-size Brazilian company (figures in Brazilian reais). The company adopted flexible hours and hybrid work, and announced it trusted the team. Six months later, leadership was complaining about deadlines, and coordinators were complaining about answering messages all day.

Both complaints had the same root. The company had removed presence control without putting anything in its place: nobody knew what the priority was that week, or who decided what to do when two clients asked for the same thing. Autonomy without context is not freedom; it is a person guessing alone and then being judged for it.

01The idea: whoever does the work knows more about it #

Peter Drucker published Managing Oneself in the Harvard Business Review in 1999, around the same period he was writing about knowledge-worker productivity in the California Management Review. The central argument is structural, not motivational: in manual work, the method could be defined by someone outside who observed it; in knowledge work, whoever executes holds information the supervisor lacks about how the task should be done.

It follows that detailed supervision of method loses effectiveness — not for moral reasons, but because of an information asymmetry. Drucker argued that responsibility for development, allocation and contribution largely shifts to the professional, who needs to know how they learn, where they are strong and where they can actually produce results.

It is easy to slide here into corporate self-help — the version that turns an argument about work design into an exhortation for the employee to 'own their career.' That reading is convenient because it lets the company off the hook. The useful angle for a manager is the reverse one: if the person is going to decide the method, what exactly does the manager need to deliver for that decision to be possible?

02What is left for the manager — and it is not little #

Three things, and none of them is execution follow-up.

Context. What this work is for, for whom, under what constraint, what happens if it is late. Whoever has context decides well alone most of the time; whoever does not, asks — and the question eats up both people's day. Drucker already addressed this in The Practice of Management (1954), linking a defined objective to self-control.

Priority. Autonomy does not resolve conflicts between demands from different sources: that is a decision for whoever has visibility over the whole. Leaving prioritization to the person executing the work means handing them a problem only the manager can solve, and then judging them on the outcome.

Concrete feedback. Not a semiannual review: information about the effect of what was done. In The Effective Executive (1967), Drucker described the practice of writing down the expectation before a decision and comparing it afterward with what happened — the same principle applies to a team's work.

None of these three things is task follow-up, which is why they tend to fall out of a manager's routine: follow-up is urgent and visible, while context, priority and feedback are important and quiet. The result is a common inversion — the manager spends the day answering questions about other people's work and never finds time to write the thing that would make those questions disappear.

There is also a practical difference between delegating a task and delegating a decision. Delegating a task keeps the dependency alive: the person executes and comes back to ask what's next. Delegating a decision requires stating in advance the criterion, the limit and what to do in the ambiguous case. It takes more effort the first time and almost none afterward, because the criterion holds for the whole class of similar situations.

One warning about the most common side effect. When context is not available, the most experienced people become the informal source — and end up interrupted in the manager's place. The operation seems to work, but it rests on two or three people who lost their own agenda to sustain everyone else's. They rarely complain; they usually resign.

03The cost of the question that should not exist #

Want to see how this looks inside a real operation? Explore the platform.

Scale of the 130-person firm, with stated assumptions to rework with your own numbers:

The math: 48 minutes a day × 220 working days gives about 176 hours per person a year; across 130 people, 22,900 hours; at a loaded R$ 80 an hour, roughly R$ 1.8 to 2.1 million. It is a deliberately rough estimate — even if you cut it in half, what's left is the annual cost of an entire team spent asking what could already have been written down.

04Where this gets stuck in practice #

1. Autonomy is announced, not installed. Control is removed and no context is put in its place. The team reads it as abandonment, not trust.
2. Priority lives in the manager's head. Until it is written somewhere searchable, it will be requested by message, one person at a time.
3. Context is given once, in the kickoff meeting. Whoever joins the work later never receives it.
4. Feedback only exists when something goes wrong. Without information about what went well, the person has no way to calibrate their own judgment.
5. Availability gets confused with performance. Answering fast in chat becomes the visible indicator, and deep work has nowhere to sit.

05What management with data answers #

The practical condition for autonomy is mundane: the state of the work needs to exist outside of conversations. Written priority, an explicit owner, visible progress, decisions logged with the reason. With that, the manager stops being the information router — a role that eats their day and still slows everyone else down.

It is the direct counterpart to what appears in the manager's fragmented day and in the cost of coordinating. The link to a declared, verified objective is in from MBO to OKR, and the manager behavior that sustains it, in what makes a good manager, in data.

On the platform, this is the tasks, projects and people modules, with priority and ownership in the same place as the work. Solutions by area show the specific cut, and a conversation helps spot where the biggest loss is today.

If your team has to ask what to do first, the problem is not a lack of autonomy. It's a lack of available context — and that part is yours.