You Probably Don't Have a Headcount Problem
Work is piling up and the obvious answer is another person. Before you open the role, find out whether the queue is waiting on capacity or on a decision nobody owns. Those need opposite fixes.
The backlog is growing. Deadlines are slipping by a few days at a time, which is worse than slipping by a week because nobody escalates it. Your best people say they are underwater and you believe them, because they are.
So you open a role. It is the one lever that feels available, it is socially easy to justify, and everybody in the room agrees the team is stretched.
Six months later the person is hired, onboarded, and busy. The backlog is roughly the same size.
What capacity actually looks like when it is the problem
A genuine capacity shortfall has a specific shape. Work arrives, someone picks it up immediately, moves it forward without stopping, and finishes it. The only reason more does not get done is that there are not enough hours in which to do it. Everyone is busy on the work itself, not on chasing it.
That is rarer than it sounds. In most operations that feel short-staffed, people are not saturated by the work. They are saturated by the overhead around the work: following up, re-explaining, waiting, checking whether something was ever decided, redoing a piece that came back.
Adding a person to a genuine capacity problem adds capacity. Adding a person to a coordination problem adds coordination. The second one gets worse with headcount, not better, because every additional person is another party to every handoff and another context to keep current.
Follow one piece of work
The fastest way to tell which you have is to stop reasoning about the category and look at one instance.
Take a single real piece of work that took too long. Not a typical one, an actual one, last month. Reconstruct its life from the moment it entered the business to the moment it was genuinely finished. Write down every point at which it moved and every point at which it stopped.
Then add up the time. Not the total elapsed time, the two columns separately: time when somebody was actively working on it, and time when it was sitting.
In an operation that truly needs people, most of the elapsed time is in the first column. In most operations that feel like they need people, the great majority is in the second, and the sitting is not idleness. It is work waiting on something.
Waiting on a decision and being worked on look identical in every status report ever written. Both show as in progress. That is why an operation can be mostly queue and still report as busy, and why the person who feels it most is the one who keeps having to ask.
What the sitting is usually made of
Three things, in roughly this order.
Decisions nobody owns. Something needs a call. It is not obvious whose call it is, so it waits, or it goes to the most senior person available, who is the least available person. Nobody experiences this as a delay they caused. Everybody experiences it as someone else being slow.
Rework, because finished was never defined. Work is handed on when the person stops touching it. The next person finds it incomplete by their own standard, and back it goes. Two people were busy. Nothing moved. Hire a third and you have three people doing this.
Exceptions running on judgment. The standard path does not cover every case, and the cases it misses get handled by whoever notices, using experience, off the record. That works, right up until the volume of exceptions is high enough that your most capable person is spending their week on them, which is the same week you wanted them to spend on the backlog.
None of these three is solved by another pair of hands. All three are made slightly worse by one.
The test, and what to do about each answer
Run the exercise on one piece of work and one of two things will be true.
If the working time dominates and the waiting is short, you have a capacity problem and you should hire. You will also now know exactly what to hire for, which is not always the role you assumed.
If the waiting dominates, name what each wait was for. A wait on a decision means that decision needs an owner and, usually, needs to move down a level so it stops queuing behind someone in meetings. A wait caused by returned work means that step needs a definition of done that is an artifact rather than an opinion. A wait caused by an unusual case means that case needs to stop being unusual: give it a route, and let the route be used.
That work is unglamorous and it does not require budget approval. It also tends to return capacity faster than hiring does, because it releases capacity you are already paying for.
Hiring raises capacity. It does not create ownership, define done, or route an exception. If the work is waiting rather than being worked on, another person joins the queue instead of shortening it. Follow one real piece of work end to end, split the time into working and waiting, and let that decide.
The honest version of the answer
Sometimes you are simply short-handed, and every hour of diagnosis you spend is an hour the backlog grows. That happens.
But it is worth an afternoon to know which situation you are in, because the two look identical from the inside and respond to opposite treatments. The cost of getting it wrong is not just the salary. It is a year of believing you addressed the problem.