How to Delegate Without Losing Accountability
Kinetiq Team
In 1974, William Oncken Jr. and Donald Wass published “Management Time: Who’s Got the Monkey?” in Harvard Business Review. The metaphor was deliberately absurd: every problem is a monkey, and the monkey sits on somebody’s back. A manager walks down a corridor, someone says “have you got a minute, we have a problem,” and by the end of a four minute conversation the manager has said the words that transfer the animal. “Let me think about it and get back to you.”
The monkey changed backs. Nothing was delegated. The person now waits on the manager, the manager now owes the person an answer, and the manager’s calendar fills with obligations they acquired by accident in hallways. Oncken’s piece became one of HBR’s most reprinted articles, which tells you the problem did not get solved.
What the Research Shows
Most Failed Delegation Is a Decision-Rights Problem
Paul Rogers and Marcia Blenko’s “Who Has the D?” argued that organizational bottlenecks are usually not caused by unclear tasks. They are caused by unclear decision roles. Their framework separates who recommends, who provides input, who agrees, who decides and who performs, and the whole point of separating them is that these are different jobs that get bundled into one vague word.
“Delegate” is that vague word. When a manager says they have delegated a project, they may mean any of four things: you do the work and I decide, you do the work and recommend and I decide, you decide and tell me, or you decide and I do not need to know. Those are enormously different arrangements. The manager usually has one in mind. The other person frequently has a different one. Neither says which, because it feels obvious from where they are standing.
Accountability Does Not Move the Way Work Does
Here is the part that gets skipped. Responsibility is transferable. Accountability is not, at least not upward past you. If your team ships something broken, “I delegated that” is not a defence anyone accepts, and you would not accept it from your own manager either.
This is not a flaw in delegation. It is the definition of it. You are lending out the work and keeping the answerability. Managers who have not made peace with that arrangement tend to resolve the tension in one of two unhelpful directions: they hover, because they can feel they are still on the hook, or they perform a clean handover and then act surprised when the thing they were always accountable for goes sideways.
The Gallup Finding Cuts Both Ways
Gallup’s research on managers, which we covered in the 70% engagement variance piece, is usually read as an argument for investing in managers. It is also an argument about concentration risk. If one role explains that much of the variance in a team’s engagement, then a team whose every decision passes through that role has built something fragile around a single person’s availability, judgment and mood.
Responsibility is transferable. Accountability is not. You are lending out the work and keeping the answerability, and managers who have not made peace with that either hover or vanish.
Why This Matters for Teams
The costs show up on both ends of the spectrum, and they are easy to mistake for each other.
Under-delegation is the more visible failure. Work queues behind one person, that person feels essential, and the team learns that bringing a finished decision is riskier than bringing a question. The tell is not workload. It is what arrives in your inbox. If people bring you problems rather than recommendations, you have trained them to.
Over-delegation is quieter and lands harder. A manager hands over something genuinely consequential, sets no checkpoint because a checkpoint would feel like distrust, and finds out at the deadline. Then comes the worst possible correction: the manager takes it back, in public, at the last minute. The person learns that ownership is conditional and revocable, which is a far more durable lesson than anything said in the one-on-one afterwards.
Both failures come from the same missing piece. Not trust, and not skill. A shared understanding of who decides what, and when the manager expects to hear about it.
The Gap the Data Reveals
The research is clear that decision rights matter and that managers hoard them. What it does not supply is the thing a manager needs on a Tuesday, which is a sentence to say.
Most delegation advice stops at “be clear about expectations,” which is true and useless, roughly like telling someone to be clear about the weather. Clarity about what? Expectations are not one thing. There is the outcome, the boundary conditions, the decision authority, the reporting cadence and the definition of done, and a manager can be crystal clear about two of those and leave the other three entirely to inference.
The second gap is that delegation gets framed as a single act rather than a schedule. You do not delegate on Monday and find out on Friday. That is not delegation, that is a bet. What actually works is agreeing the checkpoints at the moment of handover, so that the check-in is a thing you both planned rather than a thing you did because you got nervous. This distinction is the entire difference between a status update and a hover, and it is set at the start or not at all.
How to Delegate Without Dropping the Thread
Say Which of the Four It Is
At handover, name the arrangement out loud. “You decide, tell me after.” “You recommend, I decide by Thursday.” “You own this completely and I do not need to see it.” Four seconds. It removes the single largest source of delegation failure, which is two people holding different models of who has the D and both being certain.
The wording matters less than the fact that you said it. What you are ruling out is the silent mismatch where you thought you had handed over a decision and they thought they were doing research for you.
Set the Checkpoint on Inputs, Not Outputs
The instinct is to check the work. The more useful checkpoint is earlier, on the approach. Ask to see the plan, the shortlist, the first draft of the reasoning. If the approach is sound, the output usually is. If the approach is wrong, you have found out at a point where correcting it costs an hour rather than a fortnight.
This also solves the trust optics. Reviewing an approach reads as engagement. Reviewing a finished deliverable reads as inspection, because by then the only available verdict is a verdict on them.
Refuse the Monkey Politely
Oncken’s practical advice still works and almost nobody follows it. When someone brings you a problem, do not accept the next move. “What are you going to do?” is the whole technique. If they do not have an answer yet, the next move is theirs, which is to come back with one. You can absolutely help them get there. What you should not do is leave that conversation owing them something.
The exception is real and worth naming: sometimes the next move genuinely is yours, because it needs authority or information they do not have. Say so explicitly, and give a date. An accepted monkey with a deadline is fine. It is the undated one that piles up.
Write Down What Done Means, Before
Most disagreements about delegated work are not disagreements about quality. They are two definitions of done colliding at the deadline. Three lines is enough: what exists at the end, who has signed off, and what it explicitly does not need to include. That last clause does more work than the other two combined, because it is where scope quietly grows when nobody has bounded it.
Run the Vacation Test
Take a week off and watch what happens. Not as a metaphor. If work slows, delegation is roughly right. If it stops, the decisions were never delegated regardless of what the work looked like. If nobody notices you were gone, you may have over-corrected and become unnecessary to your own team, which is its own conversation.
Oncken’s monkeys are still in the corridor, and they still transfer in four minute conversations that neither person registers as a handover. The managers who stay on top of this are not the ones who delegate the most. They are the ones who can say, at any moment, which decisions they own, which they have lent out and when they next expect to hear about them. Everything else is hoping, and hoping does not scale past about four people.
Related Reading
Written by
Kinetiq Team
The KinetIQ editorial team. We write about the systems behind how work actually gets done: communication, decision-making, accountability, handoffs, and the execution habits that hold up when teams are under pressure.
