You hired someone and the work still comes back to you.

The work comes back because you hired a person, not a role. Tasks moved. Decisions did not. When your new coordinator has no written standard for the exceptions, every exception becomes a question, and every question routes to the one person who already knows the answer: you.

Six weeks after a clinic makes its first real administrative hire, the owner says some version of the same sentence. She is great. She is doing everything I asked. And I am still answering my phone at 8pm.

The hire is usually not the problem. The structure the hire walked into is. Here is what is actually happening, the move that makes it worse, what the break looks like inside a clinic, and the seven day check that tells you which piece is missing.

What's actually going on

A hire moves work in two layers. The visible layer is tasks: answering the phone, confirming appointments, chasing paperwork, sending the intake link. The invisible layer is decisions, meaning what to do when the task does not go the way it normally goes. Most owners transfer the first layer, keep the second, and then wonder why the phone still rings.

Gallup's research on entrepreneurs puts a number on how rare that second transfer is. Only one in four employer entrepreneurs, meaning founders with at least one employee, have high levels of what Gallup calls Delegator talent.^[1] The other three quarters are not lazy or disorganized. They are simply better at doing the work than at designing who else does it.

That gap has a price tag. In Gallup's study of 143 Inc. 500 CEOs, the ones with high Delegator talent posted an average three year growth rate of 1,751 percent, 112 percentage points ahead of the CEOs with low or limited Delegator talent, and generated 33 percent more revenue.^[1] Delegation is not a soft skill in that data. It is the growth variable.

You hired for tasks, not for an outcome

Read your own job post again. If it lists verbs (answer, schedule, file, confirm, follow up) and no results, you hired a pair of hands. Hands need direction. A person hired to own an outcome, for example every new patient reaches their first appointment without the owner touching it, has a standard to measure themselves against and a reason to solve problems instead of forwarding them.

The employee side of that data says the same thing. Globally, only one in two employees strongly agree they know what is expected of them at work, and employees who strongly agree their job description matches the work they actually do are 2.5 times more likely to be engaged.^[2] Your coordinator is not stalling. She genuinely does not know where her authority ends, so she checks.

The decisions never moved

Every role has a short list of decisions attached to it. Do we waive the late cancellation fee for this patient. Do we squeeze this person into a full Thursday. Do we refund. Do we call the pharmacy, or do we tell the patient to call. If nobody has written down who owns those, they default to whoever has always made them, which is you.

This is old and well documented ground. Bain's Paul Rogers and Marcia Blenko argued in Harvard Business Review that clear decision roles are what make an organization decisive, and that performance improves once you know where the bottlenecks sit and who is actually empowered to break through them.^[3] A decision with no named owner is not delegated. It is queued, and you are the queue.

Role clarity also turns out to be relational rather than personal. Gallup found that among employees in remote capable jobs, fewer than half strongly agreed they knew what was expected of them, but that figure jumped to between 77 and 84 percent among the people who also knew what their coworkers were expected to do.^[4] People understand their own lane once they can see the lines on both sides of it. In a two person or five person clinic, nobody has ever drawn those lines, because for years there was only one lane.

Training was shadowing, not documentation

Almost every failed handoff I look at was trained the same way. The new person sat next to the owner for a week and watched. Shadowing transfers the normal case competently. It cannot transfer the exception, because the exception did not happen during the week she was watching.

So the first real exception arrives in week three and there is nothing to consult. Gallup reports that only 12 percent of employees strongly agree their organization does a great job onboarding new people.^[5] In an owner operated clinic that number is effectively zero, because onboarding is whatever the owner had time for between patients.

Documentation does not mean a binder nobody opens. It means the five to ten exceptions that actually consume your week, written one page each, with the decision and the threshold stated plainly.

You stayed the escalation path

The last piece is the one owners resist hearing. You told her to ask you anything. You meant it kindly. What you built was a policy that routes every uncertainty back to you, and then you got frustrated when the policy worked exactly as written.

William Oncken and Donald Wass named this in Harvard Business Review and the framing has not aged: why do leaders run out of time while the people who report to them run out of work?^[6] Every time an unresolved problem crosses your desk on the way back up, you took it. You did not delegate that work. You loaned it out and accepted returns.

The move that usually makes it worse

Six months in, the owner decides the coordinator is underwater and hires a second person. This is the move that converts an annoying problem into an expensive one.

Nothing structural changed. There is still no written exception, no named decision owner, and no outcome anyone is accountable for. So now two people escalate to you instead of one, and they escalate to each other first, which adds a step and a delay. Your interruption count goes up, not down, and now you are also arbitrating between them.

What the second hire actually costs

Price it honestly. Gallup, working from Bureau of Labor Statistics turnover data, puts the cost of replacing a single employee at one half to two times that employee's annual salary, and calls that a conservative estimate.^[7] On a $45,000 coordinator, a hire that does not stick costs somewhere between $22,500 and $90,000 once you count recruiting, ramp time, the errors during ramp, and the owner hours spent covering the gap.

A hire made into an undocumented role is also more likely not to stick, which is how the same owner ends up running the same search a third time and concluding that good people are hard to find. The people were usually fine. The role was never built.

If that loop feels familiar at a level deeper than staffing, the companion piece is The Operator Trap, which covers why capable operators stay in this pattern for years and why intelligence makes it harder to see. This article is about the mechanics. That one is about why the mechanics keep getting skipped.

The break in a clinic

Here is the shape it takes. A hormone and weight management clinic, two providers, one owner running operations on top of seeing patients, hires a patient coordinator. The job post lists eleven duties. Training is four days of sitting in on calls.

Weeks one and two look like a win. Confirmations go out. The phone gets answered on the second ring. The owner tells people the hire was two years overdue.

Week three brings the first exception. A patient's labs are back, the provider wants a repeat draw before adjusting the dose, and the patient wants to know whether he is being charged twice. Nobody wrote down the answer. The coordinator texts the owner, who is mid visit.

Week four brings three more. A refill request lands two days before the patient's next appointment. The pharmacy has a question about a prior authorization. A new patient wants a Thursday slot that is already full. Three exceptions, three texts, three interruptions.

By week eight the coordinator has learned the actual rule of the job, which is that asking is the safe move. She is not failing. She is optimizing correctly for the system she was placed in. The owner is now fielding fifteen to twenty questions a day on top of a full schedule and has quietly concluded he hired the wrong person.

Meanwhile the visible metrics look great. Phone answer rate improved. Confirmation rate improved. No show rate improved. The one thing that did not improve is the only thing the owner cared about when he made the hire, which was his own attention.

What changes when the exceptions get written down

The fix is not a new hire and it is not software. It is one page per exception.

Take the repeat lab draw. Written version: repeat draws inside 30 days of the original are not billed separately, the coordinator tells the patient that directly, and it does not route anywhere. Written version of the refill question: refills inside seven days of a scheduled visit go to the visit, refills outside seven days get the standard request sent, and the coordinator escalates only if the pharmacy pushes back twice.

Neither of those took an hour to write. Both permanently removed a repeating interruption, because the decision moved with the task. That is the entire mechanic. Delegation that moves only tasks is a loan. Delegation that moves decisions is the only kind that holds.

That operating layer is what actually moved at Premier Hormone Health (2x revenue and retention), at Kingdom (telehealth from zero to multi-million in 12 months), and at Physio Plus TX (3x monthly revenue in 5 months). In all three the change that mattered was not headcount. It was who owned which decision, written down where somebody other than the owner could read it.

What to check in the next 7 days

You do not need anyone else to run this. You need a notebook and seven days. The point is not to fix anything this week. It is to find out which of the four pieces is missing.

Days 1 and 2: log every interruption

Every time someone asks you something that a written rule could have answered, write down the question and the time. Do not fix it, do not improve anything, just log. Two days is enough. The list is almost always shorter than owners expect, and the same four or five questions keep reappearing under different names.

Days 3 and 4: put one name next to each decision

Take the list and write exactly one name beside each item. Not a committee, not “check with me first.” One name. If your own name lands next to more than a third of the list, you have found your constraint, and it is not the person you hired.

Then say the threshold out loud. You decide, up to $200. You decide, unless it involves a controlled substance. You decide, and tell me after. Authority without a number attached is not authority, it is a suggestion, and people read suggestions as a reason to ask first.

Days 5 and 6: write the five exceptions that eat your week

One page each. Situation, decision, threshold, and what to do when the threshold is exceeded. Write them quickly and imperfectly. An imperfect written rule beats a perfect unwritten one every single time, because the written one can be followed while you are with a patient.

Then send them to the person who will actually use them and ask her to correct them. She has been living those exceptions for months, so she will find the gaps faster than you will, and correcting the pages is the first real act of ownership in the role.

Day 7: put the 30 and 90 day handoff tests on the calendar

The 30 day test: for one full week, your coordinator handles every item on the written list without contacting you. Count the escalations. Zero to two means the handoff took. More than that means one of the five pages is wrong or missing, and the escalation itself tells you exactly which one.

The 90 day test is harder, and it is the one that matters. A brand new exception appears, something that was never on the list. Does she write the sixth page herself and bring it to you for approval, or does she text you the question? Handling known exceptions proves the documentation worked. Producing new documentation proves the role took. That second thing is what you were actually hiring for, and almost nobody writes it into the job post.

This is the same bar covered in The Operations Standard: a process that only runs correctly while you are watching is not a process, it is a performance.

When a look, diagnostic, or embed is the next step

Most owners can run the seven day check alone, and a good number will solve this with five written pages and one uncomfortable conversation about authority. If that is you, do that, and do not pay anybody.

Three situations are where an outside look earns its keep.

The first is when you run the check and the list refuses to collapse. If the interruptions do not reduce to a handful of repeating decisions, the problem is upstream in how the work is structured, not in how it is documented. Writing pages for chaos just produces a lot of pages.

The second is when you have hired into the same seat twice and both people left. Two failures in one seat is a structural signal, not a hiring signal, and it is very hard to see from inside the seat you designed.

The third is when your escalations are clinical and administrative at the same time, which is normal in hormone, weight management, and cash based practices. Those are two different decision trees with two different owners and two different risk profiles, and separating them cleanly is not something a downloaded template handles well.

The free version is a look. You bring the interruption log and the org as it actually runs rather than as it is drawn, and we name which of the four pieces is broken. If the answer is one page and a threshold, you leave with that and nothing else happens. Apply for the diagnostic if you want a second set of eyes on the log before you hire again.

If your clinic is somewhere in the middle of this, the fuller picture sits on the clinics page, which maps the four ways a practice adds volume, hires, spend, and reporting without adding any capacity at all.

The uncomfortable summary: a hire does not create capacity. A hire creates a second place where decisions can be made, and only if you moved the decisions there. If you did not move them, you did not buy back time. You bought a person to ask you questions, and you are paying for the privilege.

Sources