How to delegate effectively: hand over the decision, not just the task

Learning how to delegate effectively starts with noticing what you kept. Most managers move the work and hold on to the decisions inside it, so the task comes back as a stream of questions and their own workload never drops. Delegation becomes real when the authority to choose moves too.

There is a moment that repeats in almost every management team we work with. A manager explains, with some exasperation, that they delegated: they handed over the task, gave the context, set the deadline. And the work still comes back three times a week as questions.

The usual conclusion is that people will not take ownership. Sometimes true. More often nothing was delegated; work was distributed.

The distinction sounds like vocabulary, and it is the only one that matters. A task contains two things: what has to be done, and what has to be chosen along the way. Move the first, keep the second, and your load stays put while theirs goes up.

TL;DR

• Delegation usually fails because the work moves and the decision stays. The person executes, but every choice inside the task routes back to the manager, so the manager's load never drops.
• Carol Leana measured delegation not as workload but as the dollar level of authority a claims adjuster held to settle a claim alone: 44 supervisors, 198 adjusters, one US insurance company.
• Delegation is conditional, not virtuous. In that study its effect on performance depended on competence on that kind of decision and on whether the person and their supervisor defined a good outcome the same way. Yukl and Fu found both conditions again thirteen years later.
• The delegation line, four levels: Do, Propose, Decide and tell, Own. The same person sits at different levels for different decisions, and moves up one decision at a time.
• Across 171 subordinate-supervisor pairs in the People's Republic of China, Chen and Aryee found delegation positively related to feeling empowered and to feeling like an insider, and those to performance and innovation.
• The level attaches to a decision, not a person, and has to be said out loud. "Use your judgement" is not level four, it is the absence of any level.

What are you actually handing over when you delegate a task?

Three separate things, and most managers move only one. The execution, meaning the work itself. The decisions, meaning the choices that surface inside the work and cannot be anticipated in a brief. And the consequence, meaning who answers for the result. Moving execution alone produces an executor, not an owner.

Take an ordinary task: a proposal for a large client. The execution is visible and fits in an email and a calendar slot. The decisions only surface as you go. What discount we accept if they push. Which delivery option leads. Whether we include the service the delivery team dislikes. Together they are the proposal.

If none has a named owner, they all come back to you, one at a time, at badly chosen moments. That is not a lack of initiative. It is rational behaviour from someone handed responsibility for an outcome without the right to produce it.

It is the architecture error we describe in the Accountability capability: the commitment gets named, and the authority that would make it possible stays a level up.

Why does "delegate more" fail as advice on how to delegate effectively?

Because delegation does not improve performance unconditionally. Carol Leana's research found that its effect on performance depended on the person's competence on that kind of decision and on whether they and their supervisor defined a good outcome the same way. We treat those as the two conditions to check before moving authority.

It is worth looking at how delegation was measured there, because the definition says more than the result. Carol Leana worked with 44 supervisors and 198 claims adjusters across 19 branch offices of a large United States insurance company, and delegation was operationalised as the dollar level of authority an adjuster held to settle a claim alone (Leana, Academy of Management Journal, 1986).

Not how many files they were given. How much they could decide alone. A 1986 study in a single sector does not describe every workplace, but that definition is the one we keep coming back to.

The useful finding: the subordinate's job competence and the congruence between their goals and their supervisor's moderated the effect of delegation on performance. The same authority produced different results depending on those conditions.

Note what the study does not say. It does not say that delegation without those conditions damages performance; a moderation result shows the effect varies, not that it reverses. It does say something uncomfortable: "delegate more" is advice with no addressee, because it never specifies to whom, on what, or under what conditions.

Thirteen years later, Gary Yukl and Ping Ping Fu found the same two conditions in a different design, using two samples of managers and subordinates plus interviews: managers delegated more to subordinates who were competent, who shared the leader's task objectives, and who had a favourable exchange relationship (Yukl and Fu, Journal of Organizational Behavior, 1999). International data. Two studies, two methods, the same two conditions.

What does the delegation line look like across four levels?

Four levels of authority on one recurring decision: Do, where the decision stays with the manager; Propose, where the person brings a recommendation; Decide and tell, where the person chooses and reports afterwards; and Own, where the person chooses and the manager sees the result at review. People move up per decision, not in general.

1. Do. The decision is mine, the work is yours. This level has a bad reputation it has not earned: it is the right setting for a new, high-stakes decision the person has never made. It turns toxic only when it stays there permanently.

2. Propose. You bring the recommendation, with the options and the reasoning; I choose. This is where judgement gets rehearsed, because the person carries it without yet carrying the risk. A working rule: if you have taken the recommended option five times out of five, they are no longer at level two, you simply have not said so.

3. Decide and tell. You choose, and you tell me the same week, in one sentence. You are not asking for approval, you are informing. This level is easy to skip, and in our experience skipping it causes much accidental micromanagement: someone moved straight from two to four drops out of sight, and everything snaps back to one.

4. Own. You choose and carry the consequence; I see the result at the agreed review. It does not mean I disappear, it means I look at the pattern rather than every instance.

The conditions Leana and Yukl and Fu both found are the gates between levels. Competence on this decision, not in general: someone excellent technically can be at level four on architecture and level one on a price negotiation, and that says nothing bad about them. And the same definition of a good outcome: if you are optimising margin and they are optimising delivery speed, transferred authority will produce decisions you then unwind.

The delegation line is not a decision map. The map we describe in team decision making is drawn once and shows who is involved. The line moves over time and shows how far one person has travelled on a single decision.

What changes for the person when they get the decision, not just the work?

Their standing in the organisation changes, not only their workload. Delegation signals, without saying so, that someone is considered competent and trustworthy. Chen and Aryee found that perception positively related to performance and innovative behaviour, which suggests transferred authority is not simply redistributed work.

Zhen Xiong Chen and Samuel Aryee studied 171 subordinate-supervisor pairs in the People's Republic of China and found delegation positively related to psychological empowerment and to perceived insider status, and those in turn to task performance and innovative behaviour (Chen and Aryee, Academy of Management Journal, 2007). It is a correlational study in a specific cultural setting, and we read it as one: it does not show that delegation produces performance, only that the two travel together.

The practical implication is direct. When you transfer a decision you also send a message about standing, and a team reads that message whether or not anyone announces it. For the same reason, pulling back a delegated decision costs more than it looks: you are not reclaiming a decision, you are withdrawing the standing you granted.

How do you set the level for a specific decision?

With three questions, asked about a decision rather than about a person. How many instances of this decision they have made, and how those went. Whether you define a good outcome the same way. And what happens if they get it wrong once. The answers give you the level, which then has to be said explicitly.

The failure pattern is worth describing, and what follows is a composite of it rather than any single client. An operations manager ends up approving every pricing exception in the team. She has said several times that people can decide for themselves, and she means it. But nobody knows up to what threshold, so everybody asks, which is rational: the cost of a question is small and the cost of an unauthorised mistake is unknown.

The fix was not a delegation workshop. It was one sentence: below a set threshold you decide and tell me in the weekly report; above it, you propose. One level, one decision, one threshold. The questions stopped with the ambiguity that produced them.

The three questions, in order:

How many instances have they made, and how did those go? Not your impression of the person, but their track record on that kind of decision. If the answer is none, the level is one or two, however senior they are.

Do we define a good outcome the same way? Ask it directly: what would you optimise here, if you could not have everything? If the answers differ, the second condition is not met.

What happens if they get it wrong once? If it is recoverable, the level can rise. If it is a lost client or a compliance problem, it stays low, which is proportionality rather than distrust.

Write the level down: a line in the person's development plan, or in the role mandate of a newly promoted manager. Without it the team assumes level one, the only level that costs nothing if you guess wrong.

What you can do on Monday

Do not restructure delegation across the team. Take the one recurring decision that comes back to you most often.

Write out the four levels and put a dot where the decision sits now, honestly, not where it ought to. Then walk the three questions with the person who would take it on, and name the next level out loud, with its threshold.

If almost every decision sits at one and two, the problem is not the people and one conversation will not fix it. That is where manager development comes in, because a management layer that does not transfer authority is an approval layer. If the pattern sounds familiar, what we ask of managers sets out what changes.

So we will leave you with one question. Of the decisions that reached you last week, how many needed your judgement, and how many arrived because nobody knew they were allowed?

FAQ

What does delegating a task actually mean?

It means transferring the authority to decide over part of the work, not only the execution. The test: if the person has to ask you before they can continue, you did not delegate the task, you distributed work and kept the decision. Carol Leana measured delegation as exactly that, decision authority rather than volume of work.

How do I delegate without losing control?

By moving the level rather than throwing the decision over the wall. The four levels (Do, Propose, Decide and tell, Own) allow a gradual transfer: at level three the person decides alone but tells you the same week, so you see the pattern without approving each instance.

Why is delegation not working in my team?

Usually because of one of the two conditions Leana identified. Either the person has not made enough decisions of that kind to have the specific competence, or the two of you do not define a good outcome the same way, so their choices look wrong even when consistent with what they optimise. A third cause is not about people: the level was never said out loud.

Who is accountable once I have delegated a decision?

Accountability for the outcome stays with the manager; authority for the choice moves to the person. The common confusion is that delegation also moves answerability to the organisation, which it does not. What moves is the right to choose, plus the duty to explain it at the agreed cadence.

Is delegating always a good thing?

Not unconditionally, and that is what most guides leave out. In Leana's study the effect of delegation on performance depended on competence on that kind of decision and on goal congruence. An unrecoverable decision handed to someone who has never made it is not brave delegation, it is a transfer of risk.

Sources

• Leana, C. R. (1986). Predictors and Consequences of Delegation. Academy of Management Journal, 29(4), 754-774. 44 supervisors and 198 claims adjusters across 19 branch offices of a United States insurance company. Delegation was operationalised as the dollar level of authority an adjuster held to settle a claim. Job competence and goal congruence moderated its effect on performance. Single sector, US, 1986.
• Yukl, G. and Fu, P. P. (1999). Determinants of Delegation and Consultation by Managers. Journal of Organizational Behavior, 20(2), 219-232. Two samples of managers and subordinates, plus interviews. Managers delegated more to subordinates who were competent, shared the leader's objectives, and had a favourable exchange relationship. International data.
• Chen, Z. X. and Aryee, S. (2007). Delegation and Employee Work Outcomes: An Examination of the Cultural Context of Mediating Processes in China. Academy of Management Journal, 50(1), 226-238. 171 subordinate-supervisor pairs in the People's Republic of China. Delegation was positively related to psychological empowerment and perceived insider status, and those to performance and innovative behaviour. Correlational.

By Szilárd Kacsó