Automate before you hire

When the back office gets busy, the instinct is to add a person. Sometimes that is right. Often it hires someone to do work that should not exist, and once the person is in the seat, the work stops being questioned. It has an owner now. It is somebody’s job.

The order matters more than the decision. Remove first, then hand over what is left, then hire for what genuinely needs a person.

Why order decides the cost

Hire first and you fix the shape of the process. The new person learns the current version, including the parts that are only there because of a tool you stopped using two years ago. A year later the process is not just manual, it is manual and defended, because somebody’s role depends on it.

Automate first and every later hire starts from a shorter list. You are recruiting for judgment instead of for keystrokes, which is a different job, usually a smaller one, and one where the person tends to stay.

Which work is worth automating

Not all of it is, and the split is more obvious than people expect once they look at the task rather than the department.

Work worth automating has a rule you could write down. If this, then that. Match the payment to the invoice. Chase the receipt after five days. Post the file to the folder named for the month. Nobody deliberates about any of it. A person doing it is a person copying a rule they already know.

Work worth a person is where the rule runs out. A supplier disputes a charge. A number looks wrong and somebody has to know whether it actually is. A customer wants an exception. You can automate the first ninety seconds of all three, which is gathering the facts, and you cannot automate the part after that.

The trap is the middle: work that looks like a rule but is really a judgment with a habit on top. Approving expenses is the common one. Everybody says the rule is the policy. Watch what happens for a month and there are four exceptions, and none of them are written down anywhere.

The test that saves the most time

Before deciding how to handle a task, ask what happens if nobody does it for a month.

Sometimes the answer is that a filing is late and there is a penalty. Sometimes it is that a report nobody reads does not get sent. That second kind is more common than anyone expects, and it does not need automating or staffing. It needs stopping.

This is the cheapest hour in the whole exercise and the one most often skipped, because deleting work feels less like progress than building something.

What automating actually costs

It is not free, and the cost is not the software.

It is writing the process down properly, which takes longer than doing the task and is the part that gets abandoned. It is deciding what happens when the rule does not fit, because something has to. And it is somebody noticing when it breaks, because it will, and an automation that fails quietly is worse than a person who forgets, since at least the person knows they forgot.

Anything automated needs an owner for exactly that reason. Fewer hours, but not zero.

Then hire, for less

By the time you get here the list is shorter and clearer. You know which work repeats, which needs judgment, and which you stopped doing entirely. That makes the role easier to write, easier to fill, and easier to cover when the person is away, because the parts that could be written down already have been.

It also means you are no longer hiring to keep up. You are hiring to own something, which is a better job and a much easier one to recruit for.