Responsibility Without Authority

Two teams with identical workloads can have entirely different outcomes, and the usual difference is how much say they have.

The finding

High demand with high control is demanding, frequently satisfying, and sustainable.

High demand with low control is the combination associated with the worst outcomes, consistently, across decades of occupational health research.

Low demand with low control is not benign either; it produces disengagement rather than exhaustion.

Control is the modifier, which makes it the cheapest available intervention in many settings.

What control means concretely

Not autonomy in the abstract. Specific, checkable things.

Schedule. When work happens, within reason.

Method. How the task is done, given a defined outcome.

Priority. Some say over what comes first when everything cannot.

Pace. Whether the rhythm is set by the person or externally imposed.

Ability to say no, or at least to say "not without dropping something else" and be heard.

Ability to fix the annoying thing without seeking approval.

Where it gets removed

Approval requirements for decisions the person is competent to make.

Monitoring that measures activity rather than outcome, which converts a professional into a machine operator.

Priorities set elsewhere and changed frequently, so that nothing the person chooses survives.

Process designed for the exceptional case and applied to every case.

Reorganisations that leave accountability in place and move authority away.

Escalation as the only route, so that nothing can be resolved locally.

Restoring it cheaply

Define outcomes, not methods. Then leave the method alone.

Raise approval thresholds. Most approval limits were set years ago and never revisited; raising them costs little and returns real autonomy.

Give teams a protected share of their own time to fix what irritates them.

Involve people in decisions that affect their work, genuinely rather than as consultation after the decision.

Let people schedule their own work where the role allows.

Remove one approval step per quarter and see whether anything breaks. Usually nothing does.

The measurement

Three questions, asked plainly, distinguish teams reliably.

Do you have enough say in how you do your work?

Can you influence what you work on?

When something in your work is broken, can you fix it?

A team answering no to the third is the clearest signal available, because it indicates people have stopped believing improvement is possible — which is the efficacy component of burnout appearing.

Where control genuinely cannot be given

Some roles have real constraints: safety-critical procedure, regulated processes, scheduled service delivery.

Say so, and give control where it exists. Shift preferences, break scheduling, method within the constraint, input on the procedure itself.

Even in the most constrained roles there is more available than is usually granted, and the difference between none and some is larger than the difference between some and more.

The approval audit

Approval requirements accumulate and are almost never reviewed. Removing them returns control at close to zero cost.

List every approval a team must obtain to do its ordinary work.

For each, find out when the threshold was set and by whom. Most were set years ago against different circumstances.

Ask what the approver actually does with the request. Frequently the answer is that they approve everything, which makes the step pure delay.

Ask what would happen if it were removed. Where the honest answer is a small increase in minor errors, weigh that against the delay and the loss of autonomy.

Remove one per quarter and watch. Almost nothing breaks, and the effect on how people describe their work is disproportionate to the change.

Control in tightly constrained roles

Where procedure is fixed, control still exists and is usually not granted.

Shift and break timing.

Which part of the process someone works on, and rotation between parts.

Pairing and who works alongside whom.

Input into the procedure itself, which is frequently the most valuable and the most withheld.

A route to flag something that does not work, with an actual response.

The difference between no control and some control is larger than between some and more, which makes this worth doing precisely where it looks hardest.