When an owner tells me delegation has not worked for them, the usual finding is that they have not tried it. They have distributed.
Distribution is taking the work currently sitting on you and splitting it across several people. Scheduling goes to whoever is free, one person answers the phone in the mornings, somebody else handles supplies and the customer complaints still come to you because they always have.
The work has moved. The problem has not. You have given each person a fragment of a process nobody owns, and now you are coordinating four partial handoffs instead of doing the work yourself. That is usually more effort, not less, which is why so many owners try it once, conclude their people cannot handle responsibility and take everything back.
The people were not the problem. Nothing was actually delegated.
The specific failure
An owner described their plan to me: everyone on the team should be able to do everything. They could do every job in the business and wanted the same from their staff, including the person they were developing into a manager.
I told them it would not work, and the reason is worth stating precisely.
Splitting a problem across several people does not make the problem smaller. You have handed one person a chunk, another person a chunk and a third person a chunk, and the underlying process is exactly as difficult as it was. It is now also fragile, because when one of those people leaves or is out for a week, the piece they were holding does not have an owner. It comes back to you.
And it compounds. A coordination problem across four people is not twice as hard as across two. In my experience it is closer to four or five times, because every additional person is another schedule, another availability window, another set of preferences and another handoff that can be dropped. Double the headcount and you have roughly quadrupled the coordination load.
At some point you cannot keep up with it, and what breaks is the customer’s experience, because everything that goes wrong anywhere else ends up in front of them.
There is a narrow case where distribution is the right move. If your team is drowning and you need three weeks of relief while you build the real solution, spreading load is a reasonable stopgap. Just name it as a stopgap. The failure mode is doing it, feeling temporary relief and calling it done.
The four things a delegated function needs
A function is delegated when all four of these exist. Miss one and the work returns to you, reliably and usually within a quarter.
1. One accountable owner
Not a group. Not “whoever is available.” One person whose job it is, who knows it is their job and who other people know to go to.
The test: when this thing goes wrong, is there a single person whose responsibility it was? If the honest answer is “well, it depends,” it is not delegated.
2. A documented process
Written down and accessible rather than held in someone’s head. This is the step owners skip most often, because documenting is slower than doing and produces nothing visible on the day you do it.
Two things it buys you. New people can be trained against something rather than by apprenticeship to you. And when the work is done badly, you have something to point at.
That second one is the underrated half. Without a written standard, holding someone accountable is one person’s memory against another’s, which is not accountability, it is an argument.
A version of this I have seen more than once: an owner has staff who perform a recurring task slightly differently from each other. Some do a particular step, some do not, and it mostly does not matter until a customer notices the inconsistency and asks why. The owner has no answer, because it was never written down and never specified. That is not a staff failure. There was no standard to fail.
3. Real authority
The person has to be able to decide things without checking with you. If every decision in their function routes back to you for approval, you have not delegated the function. You have hired an assistant for it and kept the job.
Decide in advance what they can decide. Say it explicitly. It will feel like too much authority, and some of their decisions will be worse than yours would have been. That is the cost of the hours you are buying back, and you either pay it or you keep the work.
4. A number they are measured on
The outcome, stated as something observable.
“Handle scheduling” is not a number. “You own the schedule. You are measured on how densely the week is booked and how many jobs get rescheduled inside a week” is.
Without the measure you cannot tell whether the function is working, so you supervise by watching, which puts you back inside the work you just handed off.
Tasks versus outcomes
The clearest way to see the difference:
| Task | Outcome |
|---|---|
| “Answer the phone when I’m busy” | “You own inbound. Every call gets a response the same day and you decide what gets scheduled” |
| “Help with invoicing” | “You own receivables. You are measured on days sales outstanding” |
| “Do the ordering” | “You own supply. Nothing runs out and you stay inside the monthly budget” |
| “Keep an eye on the crews” | “You own production quality. You are measured on callback rate” |
The left column leaves your plate for a week. The right column leaves permanently.
Notice the right column also implies the other three requirements. An outcome needs a single owner because a shared outcome is nobody’s. It needs authority, because you cannot hold someone to a callback rate they lack the power to influence. And it needs process, because otherwise you are holding them to a standard that exists only in your head.
What you actually get back
The obvious return is hours. The larger one, in my experience, is that the function stops being emotional.
When you own scheduling, every request is a relationship. A long-tenured employee asks for Friday off and you are weighing their circumstances, your history with them and how you will feel saying no.
When someone else owns scheduling with a real standard behind it, the same request becomes arithmetic. How many people are available, how many jobs need covering, did this request arrive with enough notice. If it did not, the answer is no, and it is not a referendum on the relationship.
That is not coldness. It is what makes consistency possible. The owner in the example above had a time-off policy of “all requests approved, just give me about a week’s notice,” entirely informal. Which meant every no was a personal exception rather than a policy, so they almost never said no, and the schedule absorbed the cost.
The fix is not to become stricter. It is to keep the generosity and add a boundary. Retain the right to decline a request, and you gain the ability to approve most of them, because approval is now a decision rather than a default you cannot afford.
What to delegate first
The order I would generally recommend:
1. Administration, before production. Administrative work is unbounded and fragments everything around it. Production work sits on a schedule and ends. Ten hours of administration costs more of your capacity than fifteen hours of production, because it is distributed across every hour of your day and you cannot think in the gaps between phone calls.
The exception: if production capacity is what is limiting your revenue, meaning you are turning away work you could sell, hire delivery first. Revenue you cannot fulfill is a different problem from time you cannot protect.
2. Scheduling, as its own function. Of everything you could hand off, this one buys back the most, and it is usually the one generating the most interruptions.
3. Collections and payment follow-up. Owners hold onto this longer than almost anything and it is rarely owner work. In most service businesses it belongs to whoever owns the customer relationship operationally.
4. Production quality and field oversight, once there are enough people to warrant it. A single person can typically oversee somewhere in the range of fifteen to twenty people across six to eight active sites, which means this layer stays small for a long time. Two people can often cover a business several times your current size.
That range is a rule of thumb from businesses I have worked with rather than a benchmark. It varies with how geographically concentrated the work is and how well-documented the standards are.
The one that has to come last
Sales, usually.
Most owner-operators are the best salesperson in the company and a lot of the close rate is personal credibility. That is real and it is worth protecting for longer than the other functions.
But the honest version is that it comes last, not never. A business where the owner is the only person who can sell has a ceiling equal to the owner’s calendar, and it cannot be sold to anyone, because you are the product.
What this costs
Every item on this page trades near-term time for eventual capacity, and the lag is the reason it keeps getting postponed.
Documenting costs time and produces nothing on the day you do it. A process that takes you twenty minutes takes an hour to write down properly. It pays back the fourth time someone uses it.
Training is expensive twice. You are paying the trainer and the trainee simultaneously, and neither is producing revenue while it happens. If a manager at $40 an hour trains someone at $26, that is $66 an hour of non-productive labor. Real money, and worth planning for rather than discovering.
Decisions will be made worse than you would make them, at least at first. If that is unacceptable to you, you are not delegating, and you should be honest with yourself about which of these you are actually choosing.
Some of it will fail. A function will get handed off, go badly and come back. That is normal and it is usually a missing-requirement problem rather than a wrong-person problem. When it happens, check the four requirements before you conclude the hire was wrong. It is almost always authority or measurement.
How to tell if it worked
Ninety days after handing off a function:
- Has the number moved in the right direction, or at least held?
- How many times did you get pulled into it?
- If that person were out for a week, would the function continue?
- Do the people who need this function know to go to them rather than to you?
If you are still being pulled in weekly, one of the four requirements is missing. In my experience it is authority more often than anything else, because authority is the one that feels riskiest to give and is easiest to quietly withhold while believing you have handed it over.
Frequently asked
How do I delegate when I cannot afford to hire? Delegation and hiring are separate. Documenting a process, moving a function to an existing employee and defining decision rights all cost time rather than payroll. Start with what you already have. Hiring becomes the question once the existing team is fully loaded.
What if my people do not want more responsibility? Some do not, and that is worth knowing early. Others have never been offered responsibility with actual authority attached and have only experienced being handed extra tasks. Those look identical from the outside and are completely different offers.
How detailed should documentation be? Detailed enough that a competent new person could follow it without asking you. Not so detailed that nobody will read it. Checklists beat manuals in most operational contexts.
Should I promote my best producer into management? Only if they can manage. Being excellent at the work and being able to hold other people to a standard are different capabilities, and promoting for the first while assuming the second is one of the more expensive mistakes an owner can make. Some of your best people should stay where they are and be paid accordingly.
What if I delegate and quality drops? Expect some drop initially. The question is whether it recovers within a quarter and whether the gap was a standards problem you had never written down. If quality drops and there was no documented standard, that is a documentation finding rather than a personnel one.
Most engagements I run spend more time on this than on anything else, because it is where the owner’s hours actually come from. If you want help working out which function to move first, here is how I work.