A lot of managers sort their task list by feel, because nobody ever gave them a method for it. The tedious or irritating tasks get handed off first. The tasks that feel like part of who you are stay where they are. That’s an understandable way to work through a long list, but it goes wrong in both directions. Some of what gets handed off should never have left the manager’s desk. But some of what stays could go tomorrow without anyone noticing a difference.
So this post offers another way to sort. It’s a short set of questions that look only at the task itself, not at whoever might end up doing it. You’ll see one manager’s month of work run through those questions, with the reason each task lands where it does. After that comes the small set of things that stay with a manager however capable the team is. This post is only about which tasks to hand over. How to hand a task over is a separate question, and so is who should get it. The posts that cover those are linked where they come up.
What should a manager delegate, and what should they keep?
A task can land in one of three places. You can delegate it outright, you can keep it, or you can delegate it with check-ins. Delegating with check-ins means the work leaves you, but you agree in advance on a few points where you’ll look at it. Which of the three is right depends on the task itself, so the questions are all about the task. Does it rely on your position as the manager? Does it carry private information about someone? What would a mistake cost? Can the know-how be passed on, and does the task keep coming back?
The person who might take the task isn’t on that list, and that’s deliberate. If the sort depends on who might take a task, you have to redo it for every name. That means the answer for a task would change every time someone joins or leaves the team. Choosing the person is still a real step, but it comes after this one. It’s Step 2 in the five steps of delegation, and the handover itself is covered in the art of delegation. There’s one place where the person does matter here, and it comes up near the end.
The best-known checklist for this question comes from HBR’s summary of Jenny Blake’s article on how to decide which tasks to delegate. If a task is “tiny, tedious, time-consuming, teachable, time-sensitive, or you’re just terrible at it,” the checklist says to hand it off. That’s a useful prompt, though it’s one practitioner’s view rather than a tested rule. The trouble is that every item on it is a reason to hand work off, so none of them tells you what to keep. The last one also depends on the manager’s own skill rather than on anything about the task. The same summary asks a better question: “What can you and only you do? How can you delegate the rest?” The five questions below try to turn that into something you can actually check a task against.
The phrase “delegation matrix” also gets used for quite different things. Some people mean a grid that plots a task’s importance against a team member’s skill. Others mean a ladder of how much authority to hand over, or charts that set out who approves what. But the matrix in this post is none of those. It’s a table with tasks down the side and the five questions across the top, and you’ll see it filled in below. If your question is how much freedom to give someone once a task is theirs, that’s what the 7 levels of delegation is for.
One more thing up front. The questions are practitioner judgment, not research findings. That’s why the last part of this post is plain about where they fall short.
Five questions to ask about a task before you decide
Sorting the list is the first half of delegation, and it’s the half that’s easiest to get right if you go in order. The questions are listed in the order you’d ask them, because the first two can end the sort on their own. They’re our editorial judgment, since we’ve tried to turn the commonly cited lists of what to delegate and what to keep into questions you can check a single task against.
The first question is whether the task runs on your position. Some tasks use authority, access or relationships that exist because you’re the manager. Signing off a settlement with a supplier is one, and asking your own boss for budget is another. Other tasks only work because they come from you. If someone is going to hear that they didn’t get the promotion they applied for, it matters who says it. You can check this by picturing someone else doing the task perfectly. If it still wouldn’t do its job, the task runs on your position. That usually happens because the other side needs to hear it from the person who holds the authority. A yes here means the task stays with you or it splits. The rest of the questions then only apply to the part that can go, which is why this question comes first.
The second question is whether the task involves information about individuals that isn’t yours to pass on. That covers pay, health and leave. It also covers performance concerns and anything someone told you in confidence. Company policy and local rules vary on exactly who may see what, so check yours. But the question here is simpler than any policy: was this information given to you because of your role? If it was, the task stays with you. The other option is that the sensitive part stays and everything around it goes.
The third question is what a wrong result would cost, and whether it could be undone. This is the part of decision making that most affects how a task gets handed over. If a mistake would be cheap and easy to fix, you can hand the task off with very little ceremony because the worst case is a quick correction. If a mistake would be costly but recoverable, the task can still go. You just place check-ins before the points where an error would become expensive. But if a mistake would be costly and hard to undo, that pushes the task toward keeping it. The other option is to split it so that the irreversible step stays with you. The idea of one-way and two-way doors in reversible decisions is a good way to think about the “can it be undone” part. The levels post linked above uses the cost of a wrong answer to decide how much authority someone gets. This post uses the same question to decide something earlier: whether the task leaves you at all.
The fourth question is whether the know-how can be written down or shown in about an hour. The hour is a rough measure, not a rule. What it tests is whether the task can leave you without you staying on as its memory. If someone can do it from a page of notes and one sit-down, they won’t need to keep coming back to ask. But sometimes the know-how is mostly judgment you’ve never said out loud. The task can still go in that case. It just needs check-ins at the moments where that judgment gets used, because those are exactly the moments the other person will need you. So this isn’t an argument for documenting everything. It’s a way of noticing which tasks carry knowledge that only lives in your head.
The fifth question is whether the task will come back. A task that recurs every week or every month repays the effort of handing it over. That’s because the explanation happens once, and the time comes back to you every time after that. A one-off task that needs a long explanation is different, since it often costs more to delegate than to do. This is the question that tends to move the tasks a manager keeps “because explaining it would take longer than doing it.” That’s true the first time, but by the fourth week it usually isn’t.
Here is how the answers fit together. A yes on either of the first two questions means keep the task, or split it and keep the part that needs you. If a wrong result would be costly and hard to undo, split the task so that step stays with you. If a wrong result would be costly but fixable, delegate with check-ins. Do the same if the know-how can’t be shown in an hour. Anything that gets through all of that gets delegated. That leaves one real gray area, which is a one-off task with a long explanation. Keeping it is a fair judgment call.
There’s one tiebreaker worth using when a task could go either way. Ask whether the task contains something worth learning. That’s a property of the task, not of any person. So when it’s there, it tips a close call toward delegating.
A manager’s task list, sorted
The list below is constructed for illustration. Colette and everyone on her team are made up, and so is the company, but the tasks are the kind that fill a first-line manager’s month.
Colette manages a customer service team of eight at a mid-sized company that supplies commercial kitchen equipment to restaurants and hotels. She reports to the head of operations. She was promoted from senior agent about eighteen months ago and still takes calls herself in the busiest hours. At the end of August she wrote down everything she’d spent time on that month and came up with fourteen items. She sorted them before thinking about who might take any of them, so the answer wouldn’t depend on whoever happened to be free that week.
First she sorted the list the way she normally would, without the questions. The shift schedule and the routine admin went onto the hand-off pile along with the first-round screening for an open role and a long-running dispute with the courier company. Mostly she put them there because she found them tedious. She already had a team lead pull the call data for reviews, so that part of the mid-year review was on the pile as well. Everything else stayed with her, including the weekly figures pack and a complaint from a key account. It also included a spreadsheet problem in the returns process that she’d been meaning to fix since her time as an agent. When she ran the five questions, five tasks changed places. Four moved toward delegating, and one moved back toward her.
Here’s the matrix. If a task splits, the later columns describe the part that can leave her. If a task stays whole, the later columns are marked n/a. That’s because the rest don’t need asking once the first question says keep.
| Task | Her position or private info? | Cost if wrong | Shown in an hour? | Comes back? | Outcome |
|---|---|---|---|---|---|
| Weekly figures pack for the head of operations | No | Low, fixable | Yes | Weekly | Delegate |
| Next month’s shift schedule | No | Low, fixable | Yes | Monthly | Delegate |
| Room and travel bookings, chasing training completions | No | Low, fixable | Yes | Weekly | Delegate |
| First-round screening for the open agent role | No | Low, fixable | Yes | Each hire | Delegate |
| Final decision on who to hire | Yes | n/a | n/a | n/a | Keep |
| First-week plan for the new hire | No | High, fixable | Yes | Each hire | Delegate with check-ins |
| Hotel group’s complaint about late deliveries | No | High, fixable | Yes | Now and then | Delegate with check-ins |
| Merging the three returns spreadsheets | No | Low, fixable | No | One-off | Delegate with check-ins |
| Dispute with the courier over failed-delivery charges | Partly | Low, fixable | Yes | Monthly | Split |
| Mid-year review for one agent | Partly | Low, fixable | Yes | Twice a year | Split |
| One-on-ones with her eight reports | Yes | n/a | n/a | n/a | Keep |
| Overtime budget request to her manager | Yes | n/a | n/a | n/a | Keep |
| A disagreement between two agents over callbacks | Yes | n/a | n/a | n/a | Keep |
| An agent’s request for reduced hours after a health issue | Yes | n/a | n/a | n/a | Keep |
The table doesn’t say why, so the rest of this section walks through the tasks where the reasoning is least obvious. The others either sorted the way anyone would expect or come up again in the next section.
The tasks that moved
The weekly figures pack is the clearest case. Colette built it herself as a senior agent, two years before she became the manager. It pulls call volumes, wait times and first-contact resolution from the phone system into a short summary for the head of operations. It takes her about two hours every Monday. She’d kept it because it felt like hers, and because the export from the phone system has a few quirks that nobody else knows about. But the questions showed that it didn’t need her at all. Nothing in it uses her authority and nothing in it is private. If a number is wrong, someone notices and it gets corrected the following week. The quirks fit on one page and one sit-down. Because the task comes back every week, the hour of explaining pays for itself almost at once. Sam Lloyd’s advice in SHRM’s piece on why managers must delegate to develop employees fits this task exactly: “Question every task, particularly those you have done for years.” Old tasks are easy to miss because they don’t feel like tasks anymore. They feel like part of the job instead.
The dispute with the courier went the other way. The courier’s drivers had marked about forty deliveries as failed, and the courier had charged for each one. Some of those charges were fair and some weren’t. Checking them means matching each charge against the driver’s notes and the customer’s call history, which is slow and dull work. Colette had been putting it off for three weeks, so she wanted to give the whole thing away. But the first question stops that, because agreeing a settlement commits the company’s money and the sign-off is hers. The courier’s account manager also deals with her directly. If she handed over the whole dispute, whoever took it would reach the point of negotiating with no authority to agree anything. The courier would also learn that the real conversation only starts once the manager is involved. That’s why the task splits. Checking the forty charges can go to someone else, along with building the list of which ones to contest, since that part passes every remaining question. The call with the account manager and the settlement stay with Colette.
Delegate with check-ins: where to look in
Three tasks landed in the middle, and for each one the useful decision was where the check-in goes. A check-in tied to a date tends to arrive either too early to tell anything or too late to fix anything. But a check-in tied to a point in the work can be placed just before a mistake would stop being cheap.
The complaint from the hotel group shows how that works. Their operations manager had written to say deliveries to four of their sites had been late three times in a month. Sorting it out means pulling order history, checking delivery records and talking to the warehouse. None of that depends on Colette’s position, because the hotel group wants the problem fixed, not a reply from a particular person. Even so, a wrong answer to a key account is expensive even when it can be repaired. That’s why the check-in goes at one point: before anything is sent to the customer. Every mistake up to that moment is internal and costs a few minutes to fix. A mistake after it is something the customer can hold the company to. If the reply needs a commitment in Colette’s name, such as a credit or a new delivery promise, that part comes back to her.
The first-week plan for a new agent starting in three weeks works the same way. It covers who the new hire sits with and which call types they hear first. It also covers when their system logins arrive and when they take their first live call. A poor first week can be recovered from, but it costs goodwill at the moment the new person is deciding what kind of place they’ve joined. So the check-in goes at the draft, before the plan is sent to the new hire. Any changes after that would look like disorganization rather than care.
The returns spreadsheets were the hardest call, and it could honestly have gone either way. The team logs returns across three spreadsheets that grew up over the years, and Colette has wanted to merge them since she was an agent. Two of the questions pulled in opposite directions. The know-how fails the fourth question, because the reasons the three sheets exist (which warehouse uses which, and how credit notes are triggered) are in Colette’s head and have never been written down. It’s also a one-off. That means the long explanation won’t be repaid by repetition, which points toward keeping it. But a bad first version can be thrown away, because nobody has to use the new sheet until it’s agreed. Because the questions disagreed, the tiebreaker settled it. Whoever does this work will end up understanding the returns process better than anyone on the team except Colette, and that’s worth having. She delegated it with a check-in at the first draft, before anyone on the team is asked to use it.
A month later
The check-in on the hotel group’s complaint caught something. Leon took the investigation, and he has been on the team for ten years. Even so, he had drafted a reply promising a two-day delivery window that the courier’s contract doesn’t cover. Because the check-in came first, it was fixed in ten minutes before the customer saw it.
The first-week plan needed no changes at all. Josie wrote it, and she’d joined only eight months earlier, so she remembered her own first week clearly. That’s why Colette told her she wouldn’t need to see the next one before it went out.
The returns work is still in draft. Agnes took it on and has asked good questions, but the first version needed two more conversations with Colette than either of them expected. Nobody knows yet whether the merged sheet will save the team time, or whether Colette would have finished it faster herself.
What should a manager never hand off?
A task stays with the manager when the manager’s presence or authority is the point of it. Difficulty has nothing to do with it, which is why plenty of hard tasks can go to someone capable while some easy ones can’t go anywhere.
There are two reasons behind that. The first is about the person on the receiving end. Someone hearing a decision about their own pay, role or conduct needs to hear it from the person who holds that authority and answers for it. If they hear it from a peer instead, even a well-briefed one, it can tell them the decision wasn’t important enough for the manager to deliver. The second reason is that accountability doesn’t move when the work does. Lloyd puts it plainly in the same SHRM piece: “Managers are still responsible for the success or failure of this person and for achieving the desired results.” UC Berkeley’s HR guide on the guiding principles of delegation says the same about the manager’s reluctance: “The reluctance to delegate tasks is understandable because you still retain accountability.”
The same Berkeley guide is where the commonly named keep list shows up. It’s a university HR guide rather than a research finding. It also frames the list from the other side, as a criterion for tasks you can delegate: “The task does not involve responsibilities traditionally associated with management (i.e., hiring, performance appraisals, disciplinary action)”. The guide gives the list but not the reason. The reason is the principle above, because each of those tasks is one where the manager’s authority is what makes it work.
That principle also tells you how to handle the cases that sit on the line, because most of them split rather than stay whole. Hiring is one. Screening applications and scheduling interviews can go, while the final decision on who to hire stays. That’s why Colette’s screening could be delegated. The criteria were written into the job ad, and anyone borderline went into a pile for her to look at, so a mistake at that stage was cheap. A performance review splits the same way. Someone else can pull together the evidence, and a team lead often does. The evidence might be call scores, attendance and customer comments. But the judgment about the rating and the conversation stay with the manager. Some companies have ratings reviewed above the manager. An HBR piece by Demeré, Sedatole and Woods explains why managers shouldn’t have the final say in performance reviews. It describes one approach: “One way to address this is with calibration committees, where senior leaders review batches of performance ratings and decide whether to adjust them.” But the rating a committee reviews still starts with the manager’s judgment, and the conversation still has to come from the manager. Our guide to performance review calibration covers how that process works. The logistics of one-on-one meetings can be handed off, but the meetings themselves can’t. A conflict between two people on your team works the same way. Someone else might help work out the facts, such as who took which callbacks and when. But the conversation with each person still stays yours.
There’s one more category, and it’s our own judgment rather than something the sources say: commitments made in your name. Colette’s overtime budget request to her own manager is one, and a promise to a key customer that they’ll hold you to personally is another. Each belongs to the person whose name is on it. So someone else can prepare the numbers, but the ask has to come from you.
Some companies also require specific sign-offs by policy, separately from all of this. Those apply whatever the principle says, so they’re worth knowing before you sort anything.
What goes wrong when managers sort their own list?
Four tasks in Colette’s list moved toward delegating: the figures pack, the hotel complaint, the returns spreadsheets and the new hire’s first-week plan. Every one of them is something she did before she was a manager. She built the figures pack as a senior agent and handled key-account complaints in that role. She knew the returns sheets from her time on the phones, and she used to sit with new hires through their first week. So she kept these tasks because they felt like hers, and because doing them herself always seemed quicker than explaining them. Lloyd lists exactly that among the reasons managers don’t delegate: “The belief that it takes less time to do the work than it takes to delegate the responsibility.” It’s often true the first time, which is why the habit lasts. The deeper reasons managers hold on are covered in the art of delegation, so they aren’t repeated here.
The opposite mistake is handing off whatever is unpleasant rather than whatever passes the questions. It often comes with a second problem, which is that the person gets the work without the authority to finish it. Lloyd separates real delegation from simply assigning tasks. Real delegation hands over responsibility for an outcome along with the authority to produce it. But he’s blunt about the second kind: “Sometimes this is merely dumping on people.” The courier dispute shows what that does to the person who gets it. They’d have been handed the slow, dull half of the work along with a negotiation they had no authority to finish. So the first time the courier pushed back, they’d have had to come back and hand it over again. People notice when they’re getting a manager’s leftovers, and over time they learn to expect them. If that pattern is already familiar, the problems of delegation looks at how to recover from it.
The list also gets old. A task that needed Colette’s judgment in August may not need it by December, once someone else has done it three times. A new hire, a reorganization or a change in what the team handles can also move tasks in or out of the keep pile without anyone deciding to move them. That’s why we suggest rerunning the sort roughly once a quarter and any time the team or the work changes noticeably. That suggestion is our own, not a source’s.
When do these questions get the answer wrong?
The five questions are practitioner judgment, not research findings. The closest things to evidence in our sources were a university HR guide, a practitioner’s checklist and an SHRM article, and none of them tested which tasks managers should delegate. So the questions are a reasoned way to sort a list, but they haven’t been proven to be the right way. There are also a few places where they clearly fall short.
The first is the person. The sort deliberately ignores who will take a task. But once a task lands in the middle, the spacing of the check-ins does depend on who has it. Josie’s first-week plan came through its check-in without a single change. Someone newer to the team would probably have needed two rounds of changes on the same plan. The person comes back in at the point where you plan the check-ins, which is why choosing the right person matters more than the questions admit.
The second is capacity. The questions can say “delegate” on a team that’s very small or already stretched, where there’s nobody with any room to take the work. Handing more over at that point is overload, not delegation. Lloyd allows for this too: “Some reasons for not delegating are legitimate.” He goes on to name the cases he means. “For example, if an organization is understaffed or managers have no one reporting to them, obviously it is very difficult to delegate responsibilities,” he writes.
Player-coaches are a related case. If most of your job is hands-on work, the questions still apply. But your keep pile will be legitimately large, and that isn’t a sign you’re doing it wrong.
Policy can also override the sort. Some organizations have a committee that sets sign-offs, hiring panels or reviews. In those places the keep pile moves with the policy, whatever the questions say.
The sort also can’t tell you whether a task should exist at all. A report nobody reads passes every question and lands neatly in “delegate,” even though the better answer would be to stop producing it.
Try it on your own list
Take your calendar and inbox from the last month and write out every task you spent real time on. Twelve to fifteen is about right. Run each one through the five questions in order, and stop at the first two if either is a yes. Some tasks will land in the middle. Decide where the check-in goes for each of those before you talk to anyone about it, and tie it to a point in the work rather than a date.
Sorting the list is the part you do alone, and it’s the easier part. The harder part is the set of conversations that follows. One is agreeing a check-in on a task without it sounding like you don’t trust the person. Another is telling someone that a task they expected to take on is staying with you. Rehearse that conversation with Merlin before you have it for real. Tell Merlin who you’re talking to and what you’re handing over. It will play the other person and push back the way they might, and it will help you find wording that sounds like you.