ManageTime

Article

Nobody reports to you: what to do with the work you are supposed to delegate

Without direct reports you cannot assign work, so delegating becomes routing: for each task that should not be yours, pick one of four exits. Trade it with a peer, ask whoever already owns the expertise, send it back to the person who requested it, or hand it to a tool. Choosing the exit is the skill, not the asking. The Eisenhower Matrix tells you to delegate quadrant three and stops there, which is sound advice for a manager and not much use to anyone else. This page covers how to pick an exit, what the research on handing work to a tool does and does not support, and what changes with the way you already work.

How do you delegate when you have no direct reports?

Quadrant three on the matrix, urgent but not important, and the instruction printed under it says delegate. Nobody reports to you. The spreadsheet still needs pulling, the form still needs filling, and the method has finished its job by telling you the task is not yours to do.

That leaves you with a routing problem rather than an authority problem, which is the better of the two, because routing works from any seat. Four exits exist. You can trade the task with a peer, taking something of theirs that suits you better. You can ask whoever already owns the expertise, which costs a favour instead of an afternoon. You can send it back to the person who asked, with a reason attached. Or you can hand it to a tool and check what comes back.

Not one of the four requires a title.

What they do require is different currencies, and that is how you choose between them. A trade spends your own capacity later in the week. An ask spends goodwill, which is finite and which most people underspend. A route-back spends a little friction with the requester. Handing work to a tool spends your review time, and it carries the risk that the output is wrong in a way you are not equipped to notice.

Two questions sort most tasks into the right exit.

First, does this work need somebody who knows your situation, your history with the client, the reason the number matters? If it does, a tool is the wrong exit and so is a stranger. Second, is there anyone whose skill at this clearly beats yours, by enough that the handover time pays for itself? When the answer to both is no, the honest conclusion is that the task is yours, and the Eisenhower Matrix has misled you about which quadrant it was in.

One thing this page treats as already settled. Whether to take the work on at all is a separate decision, made at intake, and the prioritization pillar covers the ranking that decision needs. Here the task already sits on your list, you have already judged that you should not be the one doing it, and the only open question is where it goes instead.

Should you use an AI tool or ask a colleague?

Only one of your four exits has a research literature behind it, and it is not one of the three involving people. What exists is a large body of work on whether people will take advice from an algorithm or from a human, which sits next to your question without being the same question.

The best current summary arrived in 2025.

Qin and colleagues, writing in Psychological Bulletin, pooled 442 effect sizes from 163 studies covering 82,078 participants, and argued that two judgments decide which way a person leans. One is how capable the tool looks next to a person. The other is how much the task looks like it needs personalizing. Where the tool seemed more capable and personalizing seemed unnecessary, people preferred the tool, at a pooled d = 0.27. In every other combination they preferred the person, and more strongly, at d = -0.50.

Those two dimensions are the part you can use. A recurring data pull, a format conversion, a first-pass summary of a long document: capability is plausible there, and none of it needs your history. A reply to a client who is already annoyed needs your history, and no amount of capability stands in for it. The framework will not tell you which tool to buy. It will tell you which of your tasks are even candidates.

Hold the numbers at the right distance, though.

Those studies measured what people preferred, not which choice produced better work, and the review reports the preference showing up more strongly in what participants said than in what they did. Most of the underlying studies handed people a described scenario rather than their own backlog with a deadline attached. So the two questions make a decent filter for your own reach-for-a-tool instinct, and they are not evidence that routing a task to software gets it done well.

Two older findings sharpen the picture, and both complicate the obvious reading. Logg, Minson and Moore ran six experiments, published in Organizational Behavior and Human Decision Processes in 2019, and found that lay people took advice more readily when they believed it came from an algorithm than from a person. The effect had two clear exceptions. It shrank when the choice lay between the algorithm's estimate and the participant's own, and it shrank among people with forecasting expertise. Experienced professionals leaned on algorithmic advice less than novices did, and their accuracy suffered for it.

Read that as a warning aimed at the part of your list you know best. The tasks where you are the local expert are the ones where you will discount a tool's output, and that discount is not always earned.

The other finding is about what happens once a tool gets something wrong.

Dietvorst, Simmons and Massey ran five studies for the Journal of Experimental Psychology: General in 2015 and found that people lost confidence in an algorithmic forecaster faster than in a human one after watching both make the same mistake. Participants who had seen the algorithm perform were less willing to pick it, and that held even among those who had watched it beat the human. One visible error is enough to retire a tool that was still the better option, which is worth knowing about yourself before the first bad output lands.

None of this was measured on delegation. Those participants were choosing whose advice to take, not handing a task to a colleague and then waiting on it, and none of them had a manager to explain the result to. The mechanism transfers and the numbers do not. For your three human exits there is no comparable evidence at all, which is a gap in the research rather than a verdict on the exits.

Is it acceptable to give work back to the person who asked?

Yes, and it is the exit most people skip, on the assumption that everybody else would find it rude.

Sending a task back is not a refusal. It is a statement about sequence: here is what I am already holding, here is where this would land, tell me whether that is what you want.

The wording matters less than the two things you attach to it. Give a reason drawn from your existing commitments rather than your preferences, and give a date, either the one you could do it by or the one you need an answer by. A request returned with a reason and a date reads as a plan. The same request returned bare reads as a no, and that is a different conversation, held at intake rather than here.

Trades run on the same mechanic and are easier to open, because you arrive holding something. A colleague who is quick at the thing you are slow at is usually slow at something you are quick at, and the swap becomes obvious the moment either of you says it out loud. Timing is what kills it. Offer a trade in the week you are drowning and it reads as offloading. Offer it in a quiet week and it reads as an arrangement, which is what you wanted it to be.

Then the work leaves your hands, and that is where most handoffs die.

A task you handed off is not done, and it is not yours either, so it falls out of both categories your list is built around. What works is one place for any work whose next move belongs to somebody else, reviewed at a fixed moment rather than whenever a worry surfaces. Getting Things Done calls that list Waiting For, and you can run it without adopting the rest of the method. Write down what went out, who has it, and the date you will ask again if nothing comes back. Nobody else will keep that record for you, because to everyone else the task looked handled the moment you mentioned it.

Sometimes every exit closes. The peer has no room, the requester wants it from you specifically, the expert is on leave, and the tool cannot be trusted with this one. That is not a failed routing attempt, it is information: the task is yours, it was in the wrong quadrant, and what it needs now is a slot in your week rather than another exit.

How does your working style change the answer?

Which exit you reach for first is fairly predictable once you know how you work. The four names below are this site's own, from the working-style self-check on the home page.

Architects route cleanly and late. The handoff gets documented, the reason is sound, and by the time it goes out the window for a trade has closed, so an Architect's default drifts toward the tool exit whether or not it suits the task. Sprinters barely route at all. Handing work over costs more time upfront than doing it inside a burst, so the burst wins, and the same class of task comes round again a fortnight later.

Visionaries hand off the wrong half. What leaves is the work they find dull, which is often the work that most needs their context, and what stays is the interesting piece nobody was waiting on. Improvisers route constantly and track nothing. Everything gets asked for in a corridor and agreed to warmly, then goes quiet, which makes a written record of handed-off work less optional for them than for anybody else.

Some situations this page does not describe. If your organisation runs a shared services desk or a ticket queue, that is a fifth exit with its own rules, and it beats asking a person most of the time. If the work landed on you because you are the only person who can do it, no routing exists, and the honest move is a conversation about scope rather than a search for a taker. And if a task stays with you because handing it over would feel like admitting you cannot manage your own load, the exit is still available and that feeling is not evidence.

So take the three tasks sitting in quadrant three right now and name an exit for each one, out loud, with a person or a tool attached. Two will route in under five minutes. The third is probably the one you have carried for a month, and the reason it never moved is that you were waiting for the authority to move it.

That was never going to arrive.

Frequently asked questions

How do you delegate when you have no direct reports?

You route it instead of assigning it: trade with a peer, ask whoever owns the expertise, send it back, or hand it to a tool. Each exit spends a different currency, so choose by cost. A trade spends your capacity later in the week, an ask spends goodwill, a route-back spends friction with the person who asked, and a tool spends your review time. If no exit fits, the task is yours and was never quadrant three.

Should you use an AI tool or ask a colleague?

Reach for a tool when it plausibly beats you and the work needs none of your context; ask a person when the task needs your situation understood. A 2025 meta-analysis in Psychological Bulletin pooled 163 studies and 82,078 participants. People preferred a tool where it looked more capable and personalizing looked unnecessary (d = 0.27), and a person otherwise (d = -0.50). Those studies measured stated preferences in described scenarios, not the quality of real work, so treat the two questions as a filter.

Is it acceptable to give work back to the person who asked?

Yes, provided you attach a reason drawn from your existing commitments and a date. The date is either when you could do it or when you need an answer back. A request returned with a reason and a date reads as a plan; returned bare it reads as a refusal, which belongs at intake. In practice the friction is smaller than people picture beforehand, which is most of why this exit goes unused.

How do you keep track of work after you hand it off?

Keep one list of work whose next move belongs to somebody else, and review it at a fixed moment. Each line carries a name and the date you will chase it. A handed-off task is neither done nor yours, so it drops out of both categories an ordinary to-do list is built around. Reviewing on a schedule beats checking when a worry surfaces, because the worry arrives too late to act on.

Sources