Decisions

Automate, or hire someone? It depends on the work.

This is usually presented as a cost comparison, which misses what actually decides it. The question is not what each costs — it is whether the work needs judgement, and whether it varies. Those two answers determine the right choice more reliably than any spreadsheet.

The comparison

	Automation	A person
Cost shape	Build once, low ongoing	Ongoing salary, rising
Ramp-up	Days to weeks	Weeks to months
Handles the unexpected	Badly	Well
Judgement	None	Yes
Consistency	Perfect	Variable
Scales with volume	Yes, cheaply	Requires hiring again
Represents you to customers	Poorly	Well
Fails by	Breaking visibly, or silently	Leaving, or disengaging

Automate when

The task is the same shape every time. Different details, identical process.
Volume is high and rising. This is where the cost curves diverge sharply.
Consistency matters more than judgement. Compliance records, data entry, reminders.
Nobody wants the job. Tasks people dislike get done late and badly by humans, and are usually the ones automation handles perfectly.
It happens outside working hours. Overnight, weekends, across timezones.

Hire when

The work needs judgement. Deciding, negotiating, weighing incomplete information.
It varies genuinely. Not different details — different problems each time.
A customer relationship is involved. People buy from people, particularly in trades and professional services.
Something is going wrong and you do not know why. A person can investigate. An automation can only run.
You need capability you do not have. Hiring buys expertise; automation only scales what you already know how to do.

The sequence that usually works

Not one or the other, but an order:

Automate the mechanical layer first. Data entry, chasing, reporting, scheduling.

See what remains. Often a role that was 60% administration becomes a role that is 90% the valuable part.

Then hire for that. You are recruiting for judgement rather than for processing, which is a better job and attracts better candidates.

Which is also cheaper, because one person doing only the valuable work replaces what would have been two doing a mix.

A caution about the framing. "Replace a salary with software" is the pitch, and it is usually wrong in small businesses — because the person was doing fifteen things and the automation does two. What actually happens is that their week gets better and they take on work you could not previously afford to do.

Common questions

Can AI replace an admin assistant?

Parts of the role, not the role. It handles data entry, scheduling, reminders and document processing well. It does not handle the phone call where someone is upset, the supplier who needs persuading, or the fifteen small judgement calls a day that nobody wrote down. Most businesses find the role changes rather than disappears.

Which is cheaper?

Automation, for high-volume repetitive work, by a wide margin over time. A person, for anything low-volume or judgement-dependent — because building and maintaining an automation for a task done twice a month is never worth it. Volume is the variable that decides.

What if I automate and then need to hire anyway?

That is a good outcome, not a failure. You will be hiring for a better-defined and more valuable role, and the automation continues carrying the mechanical layer underneath. The reverse sequence — hiring first, then automating — usually means hiring for a role that should not have existed.

How do I decide for a specific task?

Two questions. Does it need judgement? Does it vary genuinely, or only in details? If it needs judgement or truly varies, hire. If it is the same shape every time, automate — and be honest about which, because most people overestimate how much judgement their routine work requires.

Keep reading

AI automation for small business Workflow automation examples Book a free AI audit