Hire Someone or Automate It? Running the Numbers in NZ

An owner told me recently she was about to hire someone part time for twenty hours a week at $32 an hour to handle the phone and the paperwork. She had it costed at just over $33,000 a year. The real figure was closer to $40,000 before the person had answered a single call.

The hiring vs automation decision goes wrong at that first step far more often than at the second. Most owners compare a subscription price against a gross hourly rate, which is not a comparison at all. Once the employment costs are loaded properly, the comparison changes shape, and in a few cases it flips back the other way and makes hiring the obvious answer.

This article builds the full cost of a part-time hire in New Zealand and puts an honest automation cost next to it, then names the situations where the person wins outright.

The advertised rate is not the cost#

Take a part-time administrator at 20 hours a week. Here are the components you carry on top of gross wages. Rates change, so put your own current figures in rather than trusting numbers in a blog post.

Gross wages are the starting line. At 20 hours a week for 52 weeks, an hourly rate multiplied out gives you the base.

On top of that sits the compulsory employer KiwiSaver contribution for members who have not opted out. The minimum employer rate has sat at 3 percent for years and is legislated to step up, so check the rate that applies right now before you budget. It is calculated on gross earnings, and there is employer superannuation contribution tax on it as well.

Annual leave is four weeks under the Holidays Act, which for a part-timer means four of their normal working weeks paid at their ordinary rate. If you pay holiday pay as you go for genuinely casual work, that is 8 percent of gross earnings instead. Either way, you pay for roughly 52 weeks and receive roughly 48 worked weeks.

Public holidays add further paid days where the day falls on a day the person would otherwise work. Sick leave sits on top of that again, with an entitlement that begins after six months of employment and accumulates.

ACC levies apply to the employer as well as the earner, at a rate that depends on your industry classification. Payroll software, PAYE administration and the time somebody in your business spends running it are all real costs even when they are small.

Then there is equipment and access. A desk, a computer, a phone extension, licences for whatever software they need. Call it a one-off in year one and a smaller annual figure after that.

Recruitment and the ramp are the costs nobody budgets#

The costs above are the ones an accountant will list. The two that actually surprise people are the ones that sit in year one.

Recruitment takes time you are not currently spending. Writing the ad, paying for the listing, reading applications, shortlisting, interviewing twice, checking references, drafting the employment agreement. For most small businesses that is a working week of owner or manager time spread across a month, and owner time has a real opportunity cost because it is time not spent selling or delivering.

Then there is the ramp. A new administrator is not productive on day one, and in a small business without documented processes they are not fully productive at week six either. During that period they consume the time of the person training them as well as their own. A conservative way to think about it is that the first month or two costs you something like double the wage in combined time, and the person is producing perhaps half the output they will produce later.

Neither of those recurs, which matters. Spread across three years of employment they look modest. If the person leaves after eight months, you pay both again.

Run that table with your own numbers and you get a loaded annual cost. Divide it by the hours the person will actually work after leave, and you get a true hourly rate. For a part-time admin role in New Zealand it typically lands well above the rate on the job ad, and that is the figure any alternative should be compared against.

What the automation side actually costs#

The other column is simpler but has its own hidden items.

An AI system handling calls, website chat and DMs for a small service business generally sits in the low hundreds of dollars a month, with a setup fee at the start. Voice minutes and conversation volume can push it higher, and the difference between the quoted plan and the real bill is usually usage, so ask for the per-minute and per-conversation rates specifically. The bands are set out in what an AI chatbot costs in New Zealand.

Setup is where the effort goes. Somebody has to gather your prices, your service area, your hours and the answers to the questions you get every week, then test the system before it touches a customer. That is real hours, mostly yours, and it happens once. Budget a couple of weeks of part-time attention rather than an afternoon.

Ongoing, the system needs an owner. Someone reads a sample of conversations weekly, notices the questions it is fumbling and adds the missing answers. Half an hour a week keeps it healthy. Nobody keeps this up unless it is assigned to a person by name.

The honest comparison is not a subscription against a salary. It is a loaded employment cost against a subscription plus setup plus half an hour a week of supervision. Even loaded that way, for a role that is mostly answering repetitive enquiries, automation usually costs a fraction of the hire. The method for proving it on your own numbers is in working out the ROI of AI automation.

Where a person is clearly the better buy#

There are situations where I would tell someone to hire, and they come up more often than the automation pitch suggests.

Hire when the process is not documented. If the knowledge lives in your head and nowhere else, a system has nothing to learn from and you will spend the setup period doing the documentation you have avoided for three years. A person can be shown, ask questions and fill the gaps themselves.

Hire when the volume is low and irregular. Fifteen enquiries a week does not justify configuring and supervising a system. The setup effort will not pay back, and you are better off with someone who also does four other things.

Hire when the job needs an owner rather than an operator. Chasing a difficult debtor, or noticing that a process is broken and fixing it before anyone asks. Systems execute the process you gave them and never tell you the process is wrong.

Hire when the work needs a body in a place. Someone has to open the door and sign for the delivery.

And hire when the work is genuinely relational. High-value quotes and anything where the customer is upset go better with a person, which is the same conclusion reached in AI vs human receptionist.

The hiring vs automation split most businesses land on#

The hiring vs automation question is usually posed as either-or and rarely resolves that way. What tends to happen is that the automation absorbs the volume and the coverage, and the person keeps the judgement work.

An AI agent answers every call and message, including the ones arriving at 9pm on a Sunday, handles the routine bookings and enquiries, and passes anything unusual to a human with the conversation attached. The part-time administrator then spends their twenty hours on work that needed a person, instead of on a phone that rings all day. Businesses that go this route often find they do not need the extra hours they were about to hire, or that the hire they make can be a better one because the role is now interesting.

If you are at the point of deciding, do two things this week. Build the loaded annual cost of the hire using the table above and your own current rates. Then list every task the role would cover and mark each one as either repetitive or needing judgement. If most of the hours are repetitive, look at what an AI receptionist covers before you post the job ad. If they are not, hire, and do not let anyone talk you out of it. Local cost context for both options is in the guide to AI solutions for Auckland businesses.