An owner decides she needs administrative help. She is spending two days a week on things that are not the work, and hiring feels like the obvious answer.
She hires somebody part time. For the first three months that person asks her roughly twenty questions a day, because nothing is documented and every answer lives in the owner's head.
The interruptions get worse before they get better. Eventually it works, and it took five months to reach the point she expected in the first fortnight, because the actual problem was documentation rather than headcount.
What a person is genuinely better at
Judgement, difficult conversations, relationships, physical presence, chasing people who need chasing, and anything requiring initiative.
What automation is genuinely better at
Repetition, availability outside working hours, consistency, volume and anything where the answer is the same every time.
What most first admin roles actually contain
Mostly the second category. Answering the same questions, scheduling, sending the same documents and chasing the same things.
The step both options require
Writing the answers down. A person cannot be handed undocumented knowledge and neither can anything else, which is the sequence in deciding what to delegate first.
The cost comparison people make
Salary against subscription, which is the obvious comparison and not the important one.
The cost comparison that matters
Total cost including recruitment, training, management, cover for absence and the owner's time during the first months, against something that needs setting up once.
The volume test
Whether there is genuinely a job. Many first admin roles are half a job, which is difficult to recruit for and produces a bored employee.
The hours test
Whether the work arrives when somebody would be working. A large share of enquiries arrive in evenings and weekends, which no part time hire covers.
The judgement test
How much of the work genuinely requires a decision. Owners consistently overestimate this until they list it honestly.
The sequence that works
Document, automate the repetition, see what remains, then hire against the remainder, which is frequently a different and more senior role than the one originally imagined.
The outcome that surprises people
Many businesses that follow that sequence find the remaining work is a few hours a week, and the hire they eventually make is somebody who does the work rather than the admin.
When to hire anyway
When the work is genuinely relational, when there is a full job, or when the business needs a second person for reasons beyond the administrative load.
The question worth asking before either
What would actually stop if this work were not done for a fortnight. Some of it would cause real problems and some of it has simply accumulated as habit, and distinguishing the two usually removes a meaningful part of the perceived load.
The mistake in both directions
Hiring a person to do repetitive work wastes them, and automating something that needed judgement produces a worse result than leaving it undone. The work should be sorted before either decision, not afterwards.
Do the documentation first either way
An assistant handling the repetitive answering, the scheduling and the routine enquiries shows exactly what is left for a person to do, and the hire made afterwards is against a real job rather than a guess. The comparison is set out in ai employee vs part time admin, the handover in handing over enquiries when you hire your first employee, and the largest single category in enquiries that need a price before anything else. See it working on the demo, compare on pricing, and the sector view sits under small business.