Should You Hire Someone or Automate It? A Small Business Framework
A $65,000 hire costs $81,250 to $91,000 a year. That is the 1.25-to-1.4 rule: per TrueTools' breakdown, 7.65% employer FICA on every dollar, benefits running 20-30% of salary, and 10-20% overhead for space, equipment, software, and management time. The Bureau of Labor Statistics puts benefits at roughly 29-32% of total compensation for private-sector workers, and the average employer's share of family health coverage exceeds $16,000 a year — before the employee does a single hour of work.
Most people price the hire at the salary on the job post and the software at the subscription. The real comparison is fully loaded cost against fully loaded cost. The business that hires a person to do the invoicing before automating the invoicing is paying a $65,000 salary to avoid a $360 subscription.
Three questions sort which tasks belong to people and which belong to software.
Question 1: Is the task rules or judgment?
Automate the rules. Hire for the judgment. If the task has a clear input, a clear decision, and a clear output — invoicing, appointment reminders, data entry, follow-up emails — software will do it more reliably than a person, every time, at a fraction of the cost. If the task needs reading a situation, handling an exception, or earning someone's trust — a tricky customer conversation, a sales call, a diagnosis — that is a person's job, and automating it will cost you customers to save you payroll.
The expensive mistake is hiring a human to do robot work. A $65k employee spending half their week on data entry is a $40k-a-year data entry habit. The expensive mistake in the other direction is automating the thing customers experience when they are deciding whether to trust you.
Question 2: How much volume is there, really?
Automation pays for itself on volume. Do the honest math: how many hours a week does this task actually take, times your loaded hourly cost, times 52. A $65k employee costs you roughly $40 an hour fully loaded. Ten hours a week of invoice chasing is $20,800 a year. Invoicing software that does it for $30 a month is $360 a year. That is not a close call.
But two hours a week of a task does not justify a $5,000 automation project, and it certainly does not justify a hire. Low-volume, high-judgment work is where the contractor earns their place: pay for the hours, not the seat.
Question 3: What happens when it breaks?
Software fails silently and at scale. A person fails loudly and once. Before automating something, ask what the failure mode looks like and whether you would catch it in time. An automated email that goes to the wrong list for a week is worse than the same email never going out. Anything customer-facing needs a human checking the output until the system has proven itself — which means the first month of any automation is still a person's part-time job.
This is also the question that kills most AI automation pitches. If the task's failure mode is "a confident wrong answer about your business reaching a customer," that is not an automation candidate yet.