You may not need another hire

The signal is familiar. Deadlines slip, the team works late, and small things keep falling through. Everyone is busy, so the conclusion writes itself: we need another person. The job gets posted, and for a few weeks it feels like relief is on the way.

Payroll is the most expensive way to test that theory. A hire is a recurring cost and a permanent decision. It also costs the time of the people who are already stretched, because someone has to train the new arrival. If the diagnosis was wrong, you have added a salary and kept the problem.

The first test is simple. Describe what the new person would do all day, task by task. If the honest answer is that they would help out and take pressure off, that is not a job description. That is a process problem wearing a job title.

The second test separates volume from rework. Real volume is new work: more clients, more files, more transactions. Rework is chasing information that should have been captured once, entering the same details in two systems, and redoing files that came back incomplete. A new hire absorbs rework. It does not remove it, and next year you need another one.

The third test is where the load sits. If everyone is equally buried, you probably do need hands. If one desk is the choke point while the others wait, the real problem is that approval or knowledge is trapped there. Another body in the queue does not widen the gate.

Sometimes the answer really is a hire, and then the sequence still matters. Fix the process first, so the new person joins something written instead of inventing their own version of the job. Done in that order, the same person delivers more, because the routine half of the work is carried for them. AI can take a real share of that half: reading, drafting, checking, reminding, with a person approving whatever goes out. Count the hours the mess costs before you count the cost of a salary. Very often the cheaper fix is the one nobody posts a job ad for.