Most founders’ first instinct when drowning in admin is to search for a virtual assistant cost and see if they can afford another pair of hands. They find numbers ranging from $6 to $50 an hour depending on where they look, but that spread is often confusing. As it turns out, the hourly rate is usually the least useful figure in the entire decision-making process.
I have hired VAs and built AI systems that handle part of the same workload for a fraction of the ongoing price, without the same handover friction. Neither option is automatically right for every business.
This post breaks down the real numbers for 2026, the expenses providers rarely include in their quotes, and a straightforward way to decide which option your business actually needs.
Key Takeaways
- Virtual assistant cost in 2026 runs roughly $6 to $15 USD an hour offshore, $20 to $30 for managed agency VAs, and $35 to $60 NZD an hour for a local contractor.
- The hourly rate is about 60% of the true cost. Recruitment, training time, management overhead and rework make up the rest.
- A part-time offshore VA at 20 hours a week lands around $10,000 to $15,000 NZD a year all-in once you count your own time.
- AI handles the repeatable end of the VA job description well: inbox triage, lead response, call answering, data entry, scheduling, reporting.
- AI does not handle judgement, relationships, or anything requiring a human to physically chase someone. That is still a VA job.
- The best answer for most founder-led businesses is not either/or. Automate the repeatable work first, then hire a VA into a smaller, higher-value role.
- Hiring a VA into an undocumented business just moves the bottleneck. You become the person answering their questions instead of doing the task.
What a Virtual Assistant Actually Costs in 2026
There are three broad tiers, and the price gap between them is enormous. Knowing which tier you are shopping in saves you from comparing apples to tractors.
Offshore direct hire
Hiring directly from the Philippines, India, Latin America or Eastern Europe through platforms like Upwork or OnlineJobs.ph puts you in the $6 to $15 USD an hour range for a general admin VA. Specialists (bookkeeping, paid ads, design, technical support) sit at $15 to $30.
At 20 hours a week, that is roughly $9,000 to $12,000 NZD a year in wages. It is the cheapest route on paper and the one most founders start with.
Managed agency VAs
Agencies handle recruitment, replacement, payroll and quality control. You pay $20 to $30 USD an hour for that, sometimes more. The premium buys you a shorter time to productivity and a safety net if the person does not work out.
For a business that has already burned three months on a bad hire, that premium is usually worth it.
Local NZ or Australian VAs
A local contractor VA charges $35 to $60 NZD an hour. Same time zone, same cultural context, no language friction, able to phone your clients without it being obvious. For anything client-facing or requiring local knowledge, this is the tier that works.
At 20 hours a week, you are looking at $36,000 to $62,000 NZD a year. That is a real hire with a real salary attached, and it should be treated as one.
The quick comparison
| Tier | Hourly (indicative) | 20 hrs/week, annual (NZD) |
|---|---|---|
| Offshore direct | $6 to $15 USD | $9k to $12k |
| Managed agency | $20 to $30 USD | $17k to $26k |
| Local NZ/AU | $35 to $60 NZD | $36k to $62k |
Treat these as starting points, not quotes. Rates move with demand, currency and skill level.
The Hidden Costs Nobody Puts in the Quote
Here is where the virtual assistant cost conversation usually goes wrong. The hourly rate is the visible part. Underneath it sits a set of costs that are real, recurring, and almost never budgeted for.
Recruitment time. Writing the job ad, filtering fifty applications, running three or four interviews, doing a paid trial. Call it 10 to 15 hours of your time. At whatever your hour is worth, that is not free.
Training and ramp. A general VA needs four to eight weeks before they are genuinely useful. During that period you are still doing the work, plus explaining the work. Output goes down before it goes up. Most founders quit at week three and conclude VAs do not work.
Management overhead. This is the one that bites. Every VA needs briefing, checking, correcting and re-briefing. Two to five hours a week of your time, permanently. That is not a training cost that goes away. It is the ongoing price of having a person in the loop.
Rework. Anything ambiguous gets done wrong the first time, because the context lives in your head and not in a document. You review it, you send it back, it comes again. Budget 10 to 20% of output as rework in the first six months.
Turnover. Offshore VA turnover is high. Better offers, life changes, burnout. When they go, the knowledge goes with them, and you start the whole cycle again from the job ad.
Add it up, and the true cost is roughly 1.5 to 1.7 times the wage bill. A $12,000 offshore VA is really a $18,000 to $20,000 commitment once your own hours are priced in.
None of that makes a VA a bad decision. It makes an undocumented VA hire a bad decision. Which is the actual problem, and it is worth reading about in more detail in why admin overload is a systems problem, not a staffing problem.
What the AI Option Costs and What It Actually Does
The comparison people want to make is “VA versus AI”, as if they are two candidates for the same job. They are not. AI is very good at a specific slice of what a VA does, and useless at the rest.
Here is the honest split.
AI handles well: answering inbound calls and booking appointments, responding to new leads within seconds, triaging and drafting email replies, chasing quotes and unpaid invoices on a schedule, entering data between systems, generating a daily summary of what happened in the business, reactivating dormant contacts in a database, first-line customer questions.
AI handles badly: anything requiring judgement about a specific relationship, negotiating, physical world tasks, work where the cost of being wrong is high and there is no review step, and anything you cannot describe clearly enough to write down.
The cost side is where the shape differs most. A VA costs the same every month regardless of volume. AI has a setup cost and then a running cost that barely moves with volume. Doubling your call volume roughly doubles a VA’s hours. It does not double the cost of an AI answering the phone.
On raw tooling, the numbers are small. ChatGPT Pro is around $20 USD a month. Claude sits in the same range. Voice AI providers charge per minute of call time, usually cents. The genuine cost of an AI setup is not the software. It is the build: getting the context of your business into a form the AI can read, connecting it to your CRM and calendar, and testing it until it behaves.
That build is a one-off, and once it is done it does not resign, does not need a briefing, and does not forget what you told it in March.
For a worked example of what that looks like in practice, one dental client, Dr Claire, was losing close to half her inbound calls despite having two receptionists on the desk. Missed calls went to zero after we put AI on the phones, and booked appointments went up 44%. No new headcount. That is the shape of the win: not replacing the team, removing the thing the team could never keep up with.
The 78% of deals that go to the first business to respond is the same story from a different angle. Speed is a machine problem, not a people problem.
Where a VA Still Wins
I am not going to pretend AI is the answer to everything, because that is how founders end up with an expensive build sitting unused.
Hire a person when the work is genuinely varied and changes week to week. AI does defined jobs brilliantly and undefined jobs badly. If you cannot write down what you want done, a VA will figure it out and an AI will not.
Hire a person when the work needs to be chased. Following up with a supplier who is ignoring you, sorting out a shipping problem, getting a straight answer out of a council. That is human work.
Hire a person when relationships are the product. Some clients want a name and a voice they recognise. If your business runs on that, do not automate it away to save $600 a month.
Hire a person when the volume is genuinely low. If you get four enquiries a week, building an automated response system is over-engineering. Answer the phone.
How to Decide: The Two-Column Test
This takes about twenty minutes, and it is the single most useful thing you can do before spending money either way.
Open a document. List every task you were going to hand to a VA. Be specific. Not “admin”, but “reply to enquiry emails”, “book jobs into the calendar”, “send the Friday invoice reminders”, “update the CRM after each call”.
Now split the list into two columns.
Column A: repeatable. Same trigger, same steps, same output, every time. Someone new could do it correctly from a written instruction. Email triage, appointment booking, invoice chasing, data entry, lead response, report generation.
Column B: judgement. Different every time. Requires knowing this specific client, or this specific situation, or reading the room.
Count the hours in each column.
If Column A is most of the list, you do not have a staffing problem. You have an automation problem, and hiring a person to do machine work is the expensive way to solve it. Automate Column A, then decide whether Column B is big enough to justify a hire at all.
If Column B is most of the list, hire the VA. But document Column B properly first, because otherwise you become the bottleneck feeding them context all day and you have bought yourself a new job.
Most founders I run this with find Column A is 70 to 80% of what they were about to hire for. Which means the VA they were about to bring on would have spent four days a week doing work a system could do, and one day a week doing the work they actually needed a human for. That is a bad trade at any hourly rate.
If you want to go deeper on how automation gets priced when you go that route, what AI automation actually costs breaks down the models, and AI agents for small business covers what these systems can genuinely do day to day.
The Real Question Behind the Cost Question
When a founder starts looking for help, they are almost never just shopping for a person. They are actually looking for a way to stop being the one who handles everything. Hiring is simply the first solution that comes to mind because it is what businesses have traditionally done when the hours run out.
The problem is that adding a person to a business with no documented processes often just shifts the load. Instead of completing the task, you find yourself researching a virtual assistant cost, explaining manual steps, and answering endless questions. This is why many founders feel even busier six weeks after hiring—the load didn’t disappear; it just changed shape.
A better order is to document your knowledge, automate the repeatable work, and then hire only for the human tasks that remain. This ensures the role is better defined and the person you hire becomes productive in days because the necessary context finally exists outside your head.
If you want a straight read on which of your tasks are Column A and which are Column B, that is a twenty-minute conversation, not a project. Book a 30-minute Discovery Call, and I will go through your list with you. No pitch, no slides. If the answer is hire a VA, I will tell you to hire a VA.
Frequently Asked Questions
How much does a virtual assistant cost per hour in 2026?
It depends entirely on where they are based. Offshore direct hires from the Philippines or Latin America run $6 to $15 USD an hour for general admin. Managed agency VAs cost $20 to $30 USD. A local New Zealand or Australian contractor charges $35 to $60 NZD an hour. Specialist skills like bookkeeping or paid ads sit at the higher end of each band.
Is a virtual assistant cheaper than an employee?
Usually yes on paper. You avoid holiday pay, KiwiSaver or superannuation, ACC levies, equipment and office costs, and you only pay for hours worked. But the saving is smaller than the hourly gap suggests once you count recruitment time, training weeks, and the two to five hours a week of management every VA needs. Budget roughly 1.5 times the wage bill as the true cost.
Can AI replace a virtual assistant?
Partly. AI reliably handles the repeatable end of the job: answering calls, booking appointments, responding to leads, triaging email, entering data, chasing invoices and generating reports. It does not handle judgement calls, relationship work, or anything you cannot describe clearly enough to write down. Most businesses get the best result by automating the repeatable work and hiring a person for the rest.
How many hours a week should I hire a VA for?
Start at 10 to 20 hours. Anything less and they never build enough context to be useful. Anything more and you are committing significant money before you know whether the working relationship holds up. Run a paid two-week trial on a defined set of tasks first. If they can work from your written instructions without constant clarification, increase the hours.
What tasks should I give a virtual assistant first?
Give them the tasks you can already describe in writing, step by step, with a clear definition of done. Inbox management, calendar scheduling, data entry and basic research are the usual starting points. Anything you cannot write down yet, keep doing yourself until you can. Handing over undocumented work is how most VA relationships fail in the first month.
What are the hidden costs of hiring a virtual assistant?
Five main ones: 10 to 15 hours of your time on recruitment, four to eight weeks of reduced output during training, two to five hours a week of ongoing management, 10 to 20% rework in the early months, and turnover risk that sends you back to the start. Together these add roughly 50 to 70% on top of the quoted hourly rate.
Should I automate before hiring a VA?
If most of the work is repeatable, yes. Automating first means the person you eventually hire steps into a smaller, better-defined role with the business context already written down, so they are productive in days rather than months. Hiring first into an undocumented business tends to move the bottleneck rather than remove it, and you end up answering questions instead of doing tasks.
About Octavius
Titus Mulquiney is the founder of Octavius AI, where he builds AI brains and AI workforces for founder-led businesses stuck running everything out of their own head. Twenty years in marketing, ex-Sony product manager, ex-GM Zeal NZ. Based in Auckland, working with operators across NZ, Australia, and the US. Connect on LinkedIn.