Blog / Small Business Automation

Virtual Assistant or Automation? How to Decide What to Hand Off

By Keyona Meeks, Founder, ReRev Labs · Oct 30, 2026 · 9 min read

The short answer: Automate work that follows fixed rules, happens often, and uses information already in your tools, like data entry, reminders and reports. Hire a virtual assistant for work that needs judgment, relationships, or changes week to week. Most small businesses do best with both: automate the copy and paste first, then hire a person for the thinking.

By the end of this post, you'll be able to:

• Sort any task into "person," "system" or "both" with four questions.
• Compare published VA rates with automation costs as of October 2026.
• Write a one-page handoff doc that works as a script spec and as a VA's training guide.

"Should I hire a VA or just automate it?" tends to come up after a bad week: a full inbox, late invoices, three people waiting on a reply.

It isn't really a choice between two products. A virtual assistant and an automation are good at opposite things, and a lot of frustration comes from handing a task to the wrong one: paying a person to copy rows between spreadsheets, or asking a system to handle a client who is upset.

Here is the test I use, the real costs as of October 2026, and how the two fit together.

Should I hire a virtual assistant or automate the work?

Look at one task at a time, not your whole job. For each task, ask four questions:

1. Are the rules fixed? Could you write the steps down so that anyone, or anything, would get the same result every time?
2. Does it happen often? Daily or weekly work adds up. A task you do twice a year rarely deserves a system.
3. Is the information already in your tools? Gmail, Google Sheets, Calendar, a form, your CRM. If the inputs are already digital, a system can reach them.
4. Does it need judgment or a relationship? Reading tone, making exceptions, deciding what matters this week, being the friendly voice a client knows.

If the first three are yes and the fourth is no, automate it. If the fourth is yes, it belongs to a person. If the rules change every week, it also belongs to a person, at least until things settle down. A system built on shifting rules will be wrong more often than it is right.

This is the same filter behind what to automate first, with one extra question added for the human side.

How do a virtual assistant and automation compare?

The row I want you to notice is "what breaks." Both fail. A person leaves and takes the know-how with them. A system silently stops when someone renames a spreadsheet tab. The difference is that a system's failures are predictable and fixable, while a person's strengths are things no system can copy.

How much does a virtual assistant cost compared to automation?

Here are published numbers rather than guesses.

Virtual assistants (as of October 2026):

• A dedicated U.S.-based assistant from Time etc is listed at $390 a month for 10 hours, which is $39 an hour, dropping to $36 an hour on the 60-hour plan.
• A managed general virtual assistant from Wing Assistant is listed at $699 a month for 80 hours, or $999 a month for 160 hours. That works out to under $9 an hour.
• For comparison, an in-house hire: the U.S. Bureau of Labor Statistics puts the median annual wage for secretaries and administrative assistants at $48,310 as of May 2025, before taxes and benefits.

Automation (as of October 2026):

• Zapier's pricing page lists a free plan with 100 tasks a month, and its Professional plan starting from $19.99 a month. Costs rise with how many tasks you run.
• If your business already runs on Google Workspace, Google Apps Script comes with it, so a custom script adds no new subscription. Google's pricing page lists Business Starter at $7 per user per month. (New to it? Here is what Google Apps Script is.)

The real cost of automation is the build: your own time, or paying someone to set it up. After that, a script that runs on a timer costs close to nothing to keep running, which is why the math changes so much for frequent tasks. A VA bills for the 52nd week of data entry exactly like the first. I go deeper on build costs in how much small business automation costs.

Rather than trust a rule of thumb, run your own numbers: the payback formula in that cost post turns minutes saved per week into the months a build takes to pay for itself.

Which tasks should go to a VA, and which to automation?

Here is how I would split the tasks that come up most.

Inbox triage

Split it. A filter or script can label, sort and archive the predictable mail: receipts, newsletters, notifications. Deciding which of the remaining emails actually matter, and how to answer them, is judgment. Give that to a person.

Scheduling

Mostly automation. A booking page tied to your calendar handles the back and forth on its own. Keep a person for the exceptions: the VIP who needs a special slot, or the reschedule that needs a careful apology.

Data entry

Automate it. Copying form answers into a sheet, moving leads into a CRM, updating a status column. This is the work most likely to be done wrong by a tired human and right by a script.

Follow-ups

Automate the reminder and the first draft, keep a person on the send. The system notices the follow-up is due and puts a draft in Gmail. A human, you or your VA, reads it and presses send. That one checkpoint is the whole point. (I walk through a full build of this pattern in Why AI Agents Break in Production.)

Customer replies

Mostly a person. An automatic "we got your message" confirmation is fine. The actual answer, especially to a frustrated customer, should come from someone who can read the situation. A VA with saved reply templates is a strong setup here.

Bookkeeping prep

Split it. A script can pull receipts out of email, file them in a folder and log them in a sheet every week. Categorizing the odd transactions and chasing missing paperwork needs a person, and your accountant still makes the calls.

Why should you automate before you hire?

Because a new hire inherits whatever process you hand them. If that process is mostly copy and paste, you are paying for copy and paste. Automate the repetitive layer first, and the person you bring on spends their hours on what you hired them for: thinking, talking to clients, catching problems.

There is a second benefit. Writing the rules down for a system forces you to understand your own process. That same document becomes your VA's training guide.

How do you write a handoff doc that works for both?

One page per task. Use the same headings whether a script or a person will do the work:

1. Trigger. What starts the task? ("A new row in the inquiries sheet," "every Monday at 9am.")
2. Inputs. Where the information lives, with links. If it lives in your head, write it down here first.
3. Steps. Numbered, in order, with the exact fields or folders touched.
4. Done looks like. What exists when the task is finished: a row updated, a draft in Gmail, a file in a folder.
5. Exceptions. The cases that break the steps, and who decides each one.
6. Never without approval. Anything that sends money, signs something, or goes to a client.

Read it back. If the steps section is long and the exceptions section is short, that task is a strong candidate for automation. If the exceptions section is the long one, that is a job for a person, and the doc is their training guide.

Where does AI fit between a VA and automation?

This is where a lot of people get sold the wrong thing. You will see AI agents pitched as a VA replacement that "just handles it." I don't build that way.

Here is how we work at ReRev: AI writes the code, plain automation runs it, and AI models only show up at specific points where a person or a rule checks the result. In the follow-up example above, AI drafts. A person approves. Nothing goes to a client unsupervised. The full argument is in AI agents vs. automation.

The practical upshot: AI makes custom automation much cheaper to build, so more tasks now land on the automation side of the line. It does not make judgment cheaper. That still belongs to a person.

Where this has limits

• If your process changes every week, hire first. Automating a moving target wastes money. Let a person run it until it settles, then automate.
• If the task is rare, don't build anything. A yearly task is cheaper to just do.
• If your data isn't in digital tools yet, a system has nothing to reach. Paper, phone calls and memory need a person, or a cleanup first.
• Systems have limits too. Google Apps Script, for example, has daily quotas, such as 1,500 email recipients per day on a Workspace account. Plenty for most small businesses, not for mass email.
• Some businesses run on relationships. If your clients expect a human voice every time, that is worth paying for, and no automation should stand in for it.

If you aren't sure which side a task falls on, that is a normal place to start. Look at the real examples in our AI in the Wild library to see what others have built.

One more distinction worth knowing. If the "person" side of your list is mostly strategic, like setting priorities, running cross-team projects, or preparing for a board meeting, that is a different role than a VA. It is closer to a fractional chief of staff, one of the services we offer.

Frequently asked questions

Is automation cheaper than a virtual assistant?

For frequent, rule-based tasks, usually yes over time, because you pay mostly once to build it and very little to run it. For work that needs judgment, a VA is cheaper than trying to force a system to do it badly.

Can AI replace a virtual assistant?

Not for the parts that matter most. AI can draft, sort and summarize, but someone still needs to check the result and handle exceptions, relationships and decisions. The best use of AI is building automations faster and preparing work for a person to approve.

What tasks should I never automate?

Anything you can't undo without a person checking it first: payments, contracts and messages to clients. A system can prepare them, but a human should press the button.

Should I automate before hiring a virtual assistant?

Yes, when you can. Automating the repetitive work first means your assistant starts on judgment work instead of copy and paste, and the rules you wrote for the system double as their training guide.

How do I know if a task is ready to automate?

It happens at least weekly, the steps are the same every time, and the information already lives in a digital tool. If you can't write the steps down yet, do it by hand a few more times first.