Good delegation is four things handed over at the same time: the outcome and the standard that counts as done, the authority to make decisions within stated limits, an agreed checkpoint instead of continuous supervision, and genuine tolerance for a method that is not yours. Hand over all four and the work stays handed over. Hand over the task while quietly keeping any one of the other three and it comes back to you, usually within a month, and usually with both of you slightly worse off than before.
That is the whole model, and it explains why delegation fails in such predictable ways. It also explains why it feels harder for owners than the advice suggests. The advice treats delegation as a scheduling decision — find the task, find the person, hand it over. In practice it is a decision about authority and about trust, and those are the two things a founder has the most difficulty giving away. This article covers the four parts, how much authority to actually give, what keeping the work costs you in real numbers, and what to do when it comes back below standard.
Why does delegation keep failing?
Almost every failure is one of two things, and they sit at opposite ends of the same scale.
The first is dumping. A task is handed over with no context, no standard and no follow-up, usually at speed and usually because you are late for something else. It is not delegation, it is abdication, and it ends in the same place every time: the work comes back wrong, you fix it yourself, and you conclude that it was easier to do it yourself all along. The person on the other end draws a different conclusion, which is that they were set up to fail.
The second is micromanagement with extra steps. The task is handed over properly, and then you hover. You check in daily, ask how it is going in a tone that means show me, and override the first two decisions they make. The work technically gets done by someone else, but you have spent nearly as much time on it, the person has learned that their judgement will not be used, and next time they will ask you rather than decide. You have created a new interruption and called it delegation.
Both failures come from the same root, which is that only the task was handed over. The four parts below are what make the difference.
What are the four parts of a proper handover?
1. The outcome and the standard, not the method
The most important thing to delegate is the result you want. What does done look like? What does excellent look like? What does unacceptable look like? If the person can answer those three questions in their own words, they have what they need. If they cannot, no amount of process documentation will save it. Standards are best expressed as something checkable — sent by Thursday, no more than two revisions, every line item priced, the client acknowledges receipt — because an uncheckable standard becomes an argument about taste later.
2. The authority to decide
Delegation without authority is not delegation, it is task allocation with a permission queue attached. The person needs to be able to decide things inside the boundary you have set without coming back to you for each one. This is the part owners most often skip, and it is the part that decides whether you actually get your time back.
3. An agreed checkpoint, not constant check-ins
Decide up front when and how you will look at progress, then leave it alone in between. A checkpoint at the halfway mark on a two-week piece of work gives you the visibility you need without removing their autonomy. The difference between a checkpoint and a check-in is that a checkpoint is in the diary and both of you know it exists. A check-in is you being anxious at four o’clock.
4. Tolerance for a different method
They will do it differently from you. If the outcome meets the standard, the method is theirs. Intervening because you would have approached it another way undoes the other three parts in one move, because what it communicates is that the real standard was never the outcome — it was doing it your way. The only useful question is whether the result was good enough, and you agreed what good enough meant before they started.
How much authority should you actually give?
More than feels comfortable, and less than everything. The cleanest way to settle it is three buckets, written down on one page per role.
- Decide alone. Things they simply handle. No conversation, no email, no permission. This bucket should be much larger than most owners’ first draft of it.
- Flag before acting. Things they should raise with you before committing: anything above an agreed value, anything that changes a client’s price, anything with legal or safety exposure. Put actual numbers on the thresholds. “Use your judgement on the big ones” is not a threshold.
- Tell me afterwards. Things they handle and then report. This bucket exists to stop the first bucket becoming invisible, and it is what lets you keep a feel for what is happening without being in the middle of it.
Write the page, circulate it, and let people challenge it — the challenges are the useful part, because they show you where two people currently believe two different things. In most small businesses that one page removes a large share of the daily interruptions, because the majority of them are people checking something they were always allowed to decide. Our guide on how to stop being the bottleneck works through the same exercise decision by decision.
What does keeping the task actually cost?
Delegation arguments are usually won or lost on a feeling, so it is worth putting the feeling on a page.
The arithmetic below uses illustrative round numbers rather than any real business — swap in your own rates and it still works. Take a task you do every week that takes three hours — building the schedule, checking and issuing invoices, pulling the numbers together for Monday. Three hours a week across 48 working weeks is 144 hours a year. Say the person who should be doing it costs you £17 an hour once employer’s National Insurance and holiday are in the number: having them do it costs £2,448 a year.
Now value your own hour on the work you would do instead. Say £55 an hour, being the rate at which your time on pricing, quoting and client relationships produces margin. Those 144 hours are worth £7,920 if you spend them on that work — and £0 if you spend them on three more hours of a different administrative task, which is what usually happens. That is the real risk in delegation, and nobody mentions it: handing work down only pays if the returned time goes somewhere more valuable.
The handover is not free either. Expect to spend around three times the task length explaining and demonstrating it, so nine hours here, plus checking the first six runs at half an hour each. Twelve hours of your time, or £660 at the same rate. It repays in about four weeks and then keeps repaying every week the task exists. The reason owners do not do it is not that the arithmetic is difficult. It is that the twelve hours land this month and the 144 arrive in slices too thin to notice.
Delegation only pays if the time it returns goes somewhere better. Decide what you will do with the hours before you hand the task over — otherwise the work leaves your desk and the week fills up exactly as before.
How do you hand over something that only exists in your head?
Most of what an owner does has never been written down, which is why so much of it feels impossible to delegate. The way through is not a documentation project. It is four passes over the same task.
- Do it and narrate it. Next time the task comes round, do it with the person watching and say out loud why you are doing each thing. Have them write it down, not you — their notes will be in the language they will actually use.
- They do it, you watch. Same task, hands swapped. Say nothing until the end unless something is about to go badly wrong.
- They do it, you check afterwards. Review against the standard you agreed, in writing, once.
- They do it. Named date, no review, straight into the “tell me afterwards” bucket.
Four cycles, and for a weekly task that is a month. The notes they made become the process, which is the second reason this method is worth the effort: at the end you have a written procedure that did not require anyone to sit down and write a manual. Our guide on building a high-performing team covers how to turn those procedures into standards the whole team works to.
What if they do it worse than you?
They almost certainly will at first, and the honest question is whether worse is fatal or merely uncomfortable. For most tasks in most small businesses, a result at 80% of your standard, produced without you, is worth more than a result at 100% that consumes three hours of your week for the next five years. For a small number of tasks — the ones with client, safety or legal consequences — it is not, and those need the checkpoint kept in place longer.
Where the gap persists, separate two causes before you react. If the person wants to do it well and cannot yet, that is skill, and skill responds to training, examples and time. If the person can do it and does not, that is will, and it responds to a straight conversation about expectations rather than more instruction. Treating a will problem with more training is the most common wasted month in small business management, and the will and skill model is the quickest way to tell them apart before you spend it.
Why is this so much harder for owners than the advice admits?
Because most delegation problems are not skills problems. They are trust problems, identity problems or structural problems wearing a skills problem’s clothes.
Owners who built the business themselves often find it genuinely hard to believe someone else can do a thing adequately, let alone as well as they can. Sometimes that belief is accurate, in which case it is a hiring and development problem rather than a delegation problem, and it should be named as one. Often it is not accurate, and it is protecting something else: the part of your identity that is still the best technician in the building, or a quiet fear that a business which runs without you says something uncomfortable about what you are for.
The question worth sitting with is this: what would have to be true for you to trust someone else with this? The answer is almost always specific — a standard written down, three clean runs, a threshold agreed — and once it is specific it is a plan rather than a feeling. That is the conversation leadership coaching spends most of its first quarter on, because until it is answered, every other improvement in the business has to route through one person.
What to do this week
- List every task you did last week that somebody else in the business could do at 80% of your standard. Pick the one with the highest hours.
- Write the definition of done for it in three lines: done, excellent, unacceptable.
- Write the three authority buckets for the person taking it on, with real numbers on the thresholds.
- Put the four handover passes in the diary, with dates, and name the day the checkpoint stops.
- Decide now what the returned hours are for, and block them before they get absorbed.
None of that requires a new system. It requires deciding what good looks like, saying it out loud, and then leaving someone alone long enough to reach it. If you want to know whether it worked, watch what happens to the shape of your week over the following month — that is where delegation either shows up or quietly did not happen.
Get the next one in your inbox
One practical, plain-English guide for business owners each week. No spam, unsubscribe any time.
Frequently asked questions
What if I genuinely am the only person who can do the work?
Sometimes that is true, and it is worth separating the two versions of it. The first is technical: the work needs a qualification, an accreditation or twenty years of judgement nobody else in the building has. That is a hiring and training question, not a delegation one, and pretending otherwise wastes a year. The second version is far more common — you are the only person who can do it because you have never written down how it is done. That is not a fact about the work, it is a fact about the documentation. Take one task you believe is uniquely yours and write the steps down while you do it next. Write the decision rules down and it usually turns out that a good deal of what felt like judgement was actually sequence.
What do I do when the work comes back below standard?
Give it back with the standard restated, and resist the urge to fix it silently. Fixing it yourself feels generous and is the single most expensive habit in delegation: the person never learns the gap, you have now done the task twice, and next time you will trust them slightly less on no new evidence. Say what specifically fell short, against the definition of done you agreed, and give them the chance to correct it. If the same gap appears three times after clear feedback, you have stopped looking at a delegation problem and started looking at a capability or a willingness problem, and those are handled differently.
Isn’t it quicker to just do it myself?
For today, almost always. That is exactly why the habit survives. The honest comparison is not this afternoon against this afternoon, it is one hour now against fifty hours a year. Teaching a three-hour weekly task costs roughly three times the task length in explanation plus a few supervised runs, so somewhere around twelve hours of your time. It repays inside a month and then keeps repaying for as long as the task exists. The trap is that the twelve hours are visible and land this week, while the hundred and forty-four hours a year are spread so thin that they never feel like a saving. Owners who delegate well simply do that arithmetic once.
How do I delegate without losing control of quality in front of clients?
Delegate the work first and the client relationship second, with a checkpoint between them. In practice that means the person produces it, you review before it leaves the building for the first six runs, and you name the date that review stops. A time-limited check is a training tool. A permanent one is you doing the job with extra steps and a bottleneck at your desk. Set the standard in writing before the first attempt rather than after it, so the review is against an agreed definition rather than your mood on the day, and tell the client who is now handling it. Clients rarely object to a named, competent person. They object to being handed to nobody in particular.