Hiring

VA, Freelancer or Agency: Who Carries the Risk?

By Valentina Brega · 8 October 2024 · 8 min read
Three empty chairs of different kinds arranged side by side in a bare, well-lit room.
Article hero image
Three engagement models compared by who absorbs the risk when something goes wrong (16:9)

Virtual Assistant vs Freelancer vs Agency: The Difference is Risk, not Price

These three options get compared on hourly rate, which explains why so many of the resulting decisions go badly. Rate is the least informative difference between them. What separates a freelancer, a managed virtual assistant and an agency is who absorbs the consequences when the work stops: when someone quits, gets ill, underperforms or disappears mid-week.

Every engagement model is a position on that single question, and the price differences follow from it rather than causing it. Once you look at them this way the choice usually becomes obvious, because you already know how much disruption your business can survive.

The Freelancer: You Carry All the Risk

Hiring directly is the cheapest option per hour and the most expensive in every other currency. You do the sourcing, the screening, the interviewing, the onboarding, the training, the quality control and the classification paperwork. When the arrangement ends, you do all of it again from the beginning, and the institutional knowledge leaves with them. Doing that screening well is a real skill in itself, covered in how to find and vet a virtual assistant.

This is a good fit for well-defined, bounded projects where you can specify the output precisely and judge it on delivery: a one-off migration, a design job, a defined build. It is a poor fit for ongoing operational work where continuity is the entire point, because the model has no answer at all to the day your freelancer stops replying.

Grid comparing who absorbs failures across freelancer, managed assistant and agency models.Grid comparing who absorbs failures across freelancer, managed assistant and agency models.
Price is the visible difference. Who absorbs the failure is the one that matters.

The model is often dismissed unfairly: an excellent freelancer working on well-specified work is one of the best value arrangements in business. Where it goes wrong is using a project structure for operational work and then being surprised when it offers no continuity. Match the structure to the work and the model performs as designed.

The Managed Virtual Assistant: The Provider Carries Continuity

A managed engagement sits between the two extremes. You get a named assistant who learns your business and works only with you, but the recruiting, vetting, oversight and replacement all sit with the provider. You keep the working relationship and the day-to-day direction; you hand over the staffing risk.

The trade is real: you pay more per hour than hiring directly, and you accept that the provider's process sits between you and the labour market. In exchange, an assistant who leaves is the provider's problem to solve rather than a hole in your week, and someone other than you is responsible for the standard of the work.

Key takeawayAsk one question of any provider: when this person leaves, what happens on my side? The answer tells you which model you are buying, whatever the marketing says.

The other thing a managed model buys is a second opinion on the work itself. When only you and the assistant can see what is happening, quality drifts slowly and nobody notices until a client does. An account manager who reviews the engagement independently catches that drift while it is still a conversation rather than an apology, and that oversight is a large part of what separates the two middle options in practice.

The Agency: Capacity without a Face

An agency sells output rather than a person. Work is distributed across a pool, which makes it resilient to any individual's absence and easy to scale up or down quickly. For bounded, repeatable, specification-driven work this is efficient and often the right answer.

Its weakness is the mirror of its strength. Because no single person owns your account, context does not accumulate. Nobody notices that a particular client always calls on Fridays, or that one vendor needs chasing twice. For operational work where the value is that accumulated judgement, a pool model quietly underdelivers while appearing to function.

Bar chart of where business context accumulates: highest with a managed assistant who stays.Bar chart of where business context accumulates: highest with a managed assistant who stays.
Context is the asset. Ask where it lives in a year, not what it costs this month.

Virtual Assistant vs Freelancer vs Agency, Side by Side

The table below compares them on the dimensions that determine outcomes rather than the ones that appear in pricing pages. Read the risk rows first, since those are the ones you will feel.

FreelancerManaged VAAgency
Who sources and vetsYouProviderProvider
Who trainsYouProvider, then you on specificsProvider
Cover for absenceNoneProvider supplies backupPool absorbs it
If they leaveStart overProvider replacesUsually invisible to you
Context accumulationHigh, then lostHigh, and retainedLow by design
Cost per hourLowestMiddleVaries, often highest
Best forBounded projectsOngoing operational workRepeatable specified output

One more distinction worth holding onto: these models fail at different speeds. A freelancer arrangement tends to fail abruptly, the replies stop and you find out at the worst possible moment. An agency fails gradually and quietly, as context that never accumulated shows up in work that is technically correct but subtly wrong. A managed engagement fails visibly, because someone other than you is watching for it. Knowing which failure mode you can absorb is a better guide than any rate card.

How to Choose without Regretting it

Start from the work rather than the budget. If you can write a specification and judge the result on delivery, a freelancer or an agency will serve you well and the cheaper option is the better one. If the value depends on someone knowing your business, your clients, your exceptions, your standards, then continuity is the product and you should buy the model that guarantees it. Either way, work out what a virtual assistant costs across a full year before comparing headline rates.

One slice of operational work now has a fourth answer. When a task is the same steps every time, with a clear input and a result you can check, a Digital Employee can run it inside your own tools, and the continuity question changes shape: nothing resigns, but someone still has to look after it, whether that is HireTrainVA, someone on your team, or your VA. The three ways to hire sets a virtual assistant, a Digital Employee and a Hybrid Team side by side, read the same risk-first way as the table above.

Want a straight answer on which model fits?Describe the work and we will tell you which of the three suits it, including when it is not us.
Book a call →

A useful test: imagine the person you engage becomes unavailable tomorrow with no warning. Write down what happens next, hour by hour, in your business. If that account is short and untroubling, the cheaper model is fine. If it involves you personally covering client communication for an unknown number of weeks, you have just described the thing you should be buying, and the rate difference is the price of not living through it.

The expensive mistake is using a project model for operational work because the hourly rate looked better. It reads as a saving for a few months and then costs you a full re-onboarding, usually at the least convenient moment. Whichever model you choose, ask what happens when the person disappears, and take the answer seriously rather than treating it as a formality.

Free discovery call

Want this handled for you?

We map the work eating your week and match a skilled specialist, typically in 7-14 days.

Book a call →
Valentina BregaFounder of HireTrainVA · host of the Built With VAs podcast

Valentina's team recruits and manages virtual assistants for US businesses, so these playbooks come from watching where the work goes.

Frequently asked questions

Q1How quickly can someone start?
Typically 7-14 days from the scoping call to a matched assistant. They already know the common tools, so the opening week is about your process rather than about the software.
Q2What happens if the match is wrong?
That risk sits with us. Every placement is backed by a two-week replacement guarantee and overseen by a dedicated account manager, so a poor fit is our problem to solve rather than a hiring process you have to run again.
Q3Do you work outside the industries listed?
Regularly. The industry pages reflect where our bench is deepest, not the limit. Tell us what the work looks like and we will tell you whether we are the right fit.