What An Unfilled Role Actually Costs You Every Week
An unfilled role costs you three things every week: the hours of your own time (or your best people's time) that absorb its work, the work that silently stops happening because nobody owns it, and the compounding you forfeit because the seat produces nothing while empty. Most owners only count the salary they are saving. That is the one number that does not matter, because the work did not go away. It just lost its owner.
The seat can be a person or an agent
When I say unfilled role, I mean any defined bundle of recurring work with no owner. Sometimes that is a human hire you have been postponing for two quarters because the right candidate has not appeared. Increasingly, it is an agent you have not deployed: the one that would do the first-line inbox triage, the daily reporting, the follow-up sequence, the meeting notes, the proposal drafts. Either way, the role exists. The work exists. The owner does not.
This matters because the waiting tax on an undeployed agent is usually larger than on an unfilled human seat, for one reason: the agent could have been running last week. There is no recruiting pipeline. There is no notice period. The gap between "decided" and "working" is measured in days if the access is sorted, which is why the Access Audit comes first in everything I teach.
Why "we're saving the salary" is the wrong frame
Owners tell me an open role is "saving" them the cost of the hire. Look at what actually happens to that role's work:
- Some of it gets done by you, at the margin of your day, at your hourly value rather than the role's.
- Some of it gets done by your most capable people, who are now doing it instead of what you hired them for.
- Some of it gets done late, badly, or halfway, which creates cleanup work downstream.
- Some of it does not get done at all, and nobody notices for a while because there is no owner to notice.
None of those is free. The salary line stays at zero while the real cost spreads across four places where you are not looking for it.
Compute your own weekly number
I am not going to hand you a statistic. Your number comes from your inputs. Take fifteen minutes and fill in the following for one specific unfilled role.
| Input | Where to find it | Your figure |
|---|---|---|
| A. Hours per week of the role's work that currently lands on you | Last two weeks of your calendar and inbox | |
| B. Your realistic hourly value | Annual owner comp or profit divided by hours you actually work | |
| C. Hours per week absorbed by other team members | Ask them directly; they know | |
| D. Their loaded hourly cost | Payroll | |
| E. Work that is simply not happening | List it, then estimate what one week of it is worth to the business |
The weekly cost of the unfilled role is roughly (A times B) plus (C times D) plus E. Write the total down. Then multiply by the number of weeks the seat has been open. That second number is the waiting tax you have already paid on this one role.
Two notes on doing this honestly. First, B is almost always higher than owners want to admit, and A is almost always higher than they first estimate. Check the calendar instead of guessing. Second, E is the line owners skip because it is uncomfortable. The follow-up calls that never happened. The content that never shipped. The reports nobody looked at because nobody made them. Put something down even if it is a range.
The part the arithmetic misses
The table above gives you the direct weekly cost. It understates the real cost, because an empty seat produces nothing that compounds.
A filled role, human or agent, generates output in week one that makes week two easier. Recorded calls become training material. A follow-up sequence produces reply data. A reporting agent surfaces a problem you fix once rather than every month. The empty seat produces none of that, so you are not just losing this week's output. You are losing the head start on next quarter's.
This is why the waiting tax feels small in any given week and enormous in retrospect. The founders I coach who look back at an eighteen-month "we're evaluating" stretch are never upset about one week. They are upset about what the eighteen weeks of compounding would have built. The decisions that compound goes deeper on which moves have this property and which do not.
Why the seat stays empty
If the cost is this visible once you look, why does the role stay unfilled for quarters at a time? Because one of the fourteen sentences is doing the work of a decision. For human hires it is usually "I'll wait for the right person" or "I'll plan this out for the quarter first." For agents it is "I tried AI, it gave me junk," "I'm not really a tech person," or "I'll lose three days if I open that box."
Notice that none of those sentences disputes the cost. They just change the subject. That is the signature of the waiting tax: the bill is never argued with, only ignored. What is the waiting tax lists all fourteen if you want to find yours.
What to do with your number
- Split the role. List every recurring task in the seat. Mark each one as rule-based (same inputs, same steps, same output every time) or judgment-based. The rule-based tasks are agent work, and they can be running this week.
- Run the Access Audit. For the agent portion, what accounts, logins, and permissions are needed? Fix those first. Fifteen minutes. This is the step that unblocks most stalled teams.
- Deploy the agent on one task. Not the whole role. One task, with a clear definition of done, checked by a human for the first week.
- Re-scope the human hire. With the repeatable work gone, the human role is smaller, more interesting, and easier to fill. Often it is a different role than the one you were waiting to fill.
- Recompute weekly. Your number should fall every week the seat is partly filled. If it does not, you have learned something real, which is more than waiting ever taught you.
The follow-on question is whether to fill the seat now with a good-enough move or keep waiting for the right person. Hiring now vs waiting for the right person lays out that comparison as a method you can run on your own figures.
FAQ
Should I count the salary I am not paying as a saving?
No. The salary is the one cost that actually went to zero, and the work did not. Count where the work went instead: your hours, your team's hours, the cleanup from work done badly, and the work that stopped happening. Those are the real lines, and together they usually exceed the salary you think you are saving.
What if I cannot estimate the value of work that is not happening?
Write down the specific tasks anyway, then put a range on one week of them. Follow-up calls not made, content not shipped, reports not produced. A rough range is far more useful than leaving the line blank, because leaving it blank is how the seat stays empty for another quarter.
Does this apply if the role is an agent rather than a person?
It applies more, because the gap between deciding and working is days rather than months. An undeployed agent has no recruiting pipeline and no notice period. If the access is sorted, it can be running on one repeatable task this week, which makes every week of delay a choice rather than a constraint.
How do I know which parts of the role an agent can take?
List every recurring task in the seat and ask of each one: are the inputs, steps, and output the same every time? If yes, it is rule-based and belongs on an agent. If it genuinely requires judgment, keep it with a human for now. Most roles turn out to be a majority of rule-based work wrapped around a small core of judgment.