Written by the Vitre team. We analyzed 5,398 community quotes and checked the federal rules binding accounting firms that use virtual assistants.

Vitre earns a commission if you buy through some links on this page. It never changes our pick or what we write.

How To Hire The Right Accounting Virtual Assistant For Your Firm

You have seen the pitch. Cheap, pre-vetted, working inside two weeks. Then the first month arrives. One owner spent twice as long fixing duplicates as doing the books alone.

An accounting virtual assistant can work well. It rarely fails on price. It fails on systems, screening, and the disclosure rules nobody mentions.

We analyzed 5,398 owner quotes and checked the federal rules binding your firm. Below is what to fix first, how to test a candidate, which numbers hold up, and when the honest answer is no hire.

(TL;DR) What to check before hiring an accounting virtual assistant

Check your systems, test the accounting, and settle the disclosure question before you hire anyone.

Most hiring guides for this role are written by the firms selling the role. They start at step two, choosing a provider. The steps below start where the failures actually begin.

1. Get your internal systems and remote hosting working first. Documented procedures, hosted files, and a review process. Owners who skipped this report the hire failing whoever they picked.
2. Write the job down before you write the ad. The strongest screening methods all depend on knowing what the work involves.
3. Screen with a planted error, not a resume. Build a trial balance or comparative statements with deliberate mistakes. Ask the candidate to walk you through the investigation.
4. Ask one deal-breaker technical question. Unreconciled transactions inside a reconciled period is the one owners rely on most.
5. Treat certificates as software evidence only. A vendor advisor certificate proves interface familiarity. The strongest bookkeeping credential includes open-book take-home sections.
6. Use structured interviews. Write the questions first, ask every candidate the same ones, score against a rubric. Sackett and colleagues put structured interviews at .42 against .19 for unstructured.
7. Price the rework, not the rate. Owners report spending twice as long fixing a cheap hire's output as doing the work themselves.
8. Check what the route takes before you decide what you pay. One owner billing a marketplace rate found the contractor received materially less.
9. Lock down access before day one. View-only bank access, an enterprise password manager holding the second-factor seed, no shared credentials, a signed confidentiality agreement.
10. Settle disclosure before the first client file moves. Payroll status does not exempt you. An employee working outside the US still means taxpayer information is being disclosed offshore.
11. Budget review time for months, not weeks. Expect to want to fire them at month three. Month six is where successful hires turn.
12. Decide direct or agency deliberately. That single choice changes your classification exposure and your professional obligations.

Before you commit, check three things. Whether the capacity problem is really a pricing problem. Whether your state board has its own disclosure requirement. And whether your professional liability policy actually reaches a remote contractor.

The accounting virtual assistant decision table

Fix your firm's internal systems before you hire any bookkeeping help

Fix the systems first, because the internal systems decide whether the hire works at all. One accounting firm owner put the condition plainly: "without good internal systems and remote hosting, offshoring won't work".

That is the first step and the one you can start on this week. It is also the step every provider skips, because it does not involve them.

Systems here means three specific things. Documented procedures for the recurring work. Hosted files and software the assistant can reach without you in the room. And a review process you have already decided on.

If you are still emailing files and holding the method in your head, you are not ready to hire. If the work only exists as habit, the assistant will invent their own version of it.

Owners who skipped this report the same outcome. The hire failed, and the candidate got the blame. Screening, access and review all sit on top of the systems layer. Build it first and everything downstream gets easier.

Send your bookkeeping candidate a broken trial balance and watch what they catch

Send a broken file, because it is the single strongest screening method in the research. Eleven separate owners describe the same approach. Plant errors, then watch.

One owner keeps it simple: "Give them a balance sheet and income statement with a few things wrong". The candidate is not there to fix the file. They are there to say out loud how they would investigate it.

The errors owners plant are specific. Un-rolled retained earnings. A negative checking balance. Miscategorized depreciation. Comparative statements with a trial balance that does not agree.

Tax practices run the same idea on a dummy return. One firm hands over raw documents and asks the candidate to "go over the tax return with you like you are the client". That tests the accounting and the client communication together.

With only thirty minutes, plant three errors of increasing subtlety. If the candidate finds the obvious one and stops, you have your answer.

The evidence base supports this, with a caveat worth knowing. Sackett and colleagues published revised validity figures in 2022 putting work samples at .33, below structured interviews at .42. Most of the underlying work-sample studies measured people already doing the job. A good trial result is weaker evidence than it feels.

Ask what a bookkeeping candidate does with unreconciled transactions in a closed period

Ask that exact question, because it is the deal-breaker owners keep coming back to. One owner is blunt about the scoring: "If they don't know what you are talking about don't hire them".

Pair it with a second question about the journal entries for sold inventory. As one owner points out, "Software handles this transaction for you". Someone with an accounting education still knows the entries.

Between them, those two questions separate people who operate software from people who understand accounting. That distinction is the whole hiring problem in this role.

Two more red flags come up repeatedly. A candidate who cannot distinguish cash from accrual basis in conversation. And a candidate who plugs a reconciliation difference to an arbitrary account rather than finding the missing transaction.

Ask every candidate the same questions in the same order. Score them against a written rubric. Structured interviews scored .42 in the 2022 Sackett figures against .19 for unstructured ones. That is the cheapest improvement available to you and it costs an hour of preparation.

A software certificate proves your accounting candidate knows the software, not bookkeeping

A software certificate proves interface familiarity and nothing else. One certificate holder said so directly after passing: "It's really geared more towards knowledge about their software than actual accounting".

Owners keep learning this the expensive way. One hired a certified candidate expecting complex reconciliations and got someone who could not locate missing transactions. "she turned out to be way less qualified than I had hoped". Another firm hired nineteen years of stated experience and found the candidate could not roll retained earnings forward.

The bookkeeping credentials are stronger, and they are still not a substitute for testing. The AIPB Certified Bookkeeper designation requires a four-part exam, an ethics undertaking, and two years of full-time experience or 3,000 part-time hours.

Only parts one and two are proctored. Parts three and four are open-book take-home papers. The experience can be completed up to three years after the exam is passed.

The NACPB Certified Public Bookkeeper license requires three certification exams and one year, 2,000 hours, of supervised bookkeeping experience. NACPB also allows experience gained inside its own program to substitute for part of that. Neither body requires a degree.

A bookkeeping credential is evidence a candidate has studied. A vendor advisor certificate is evidence they can navigate a menu. Either way, still plant the errors.

Cheap accounting virtual assistants get expensive once you count the rework

Cheap gets expensive because rework is not on the invoice. Across the 5,398 community quotes in our research base, this is the most consistent complaint at the low end of the market.

One owner reports spending twice as long recategorizing transactions and fixing duplicates as doing the books alone. Another sets the floor of realistic expectation: "I wouldn't expect a $20/hr candidate to handle anything independently". A third puts it in four words: "you get what you pay for".

The real cost has four parts. The hourly rate. The extra software seats and secure hardware. Your own unbilled hours reversing categorization errors. And the client relationship if a filing goes out wrong.

Now the independent anchors, because they are the only figures here that are not sold to you. The Bureau of Labor Statistics puts the median for US bookkeeping, accounting and auditing clerks at $23.66 an hour, or $49,210 a year. The bottom ten percent earn under $34,600 and the top ten percent over $72,660.

The Philippine Statistics Authority, in its 2022 wages survey, puts accounting and bookkeeping clerks at PHP 19,676 a month. At roughly 57 pesos to the dollar that is about $345, though treat the conversion as approximate arithmetic rather than a published figure.

One owner in the research reports paying $4 an hour. Against the Philippine data that is roughly double the local benchmark, so it is a plausible floor. It is one firm's experience, not a market rate.

Two comparisons to refuse. A monthly band of $2,500 to $4,000 matches published seller pricing almost exactly, so treat it as what agencies charge rather than what the work costs. And any figure near $300 an hour is professional billing, sitting about eight times above the BLS ninetieth percentile.

Watch what the marketplace takes before you decide what your bookkeeper costs

Watch the deductions, because the rate you post is not the rate the person receives. This is the arithmetic no seller runs for you.

One owner hiring at a marketplace rate of $45 an hour notes that "the contractor is making less than that" after the platform's cut. Coming the other way, an owner using an offshore arrangement lands at a different number than expected: "I'm paying about $12/hour, which isn't much cheaper than US".

That second figure is the one that should change decisions. If your all-in offshore cost is $12 an hour against a US median of $23.66, the gap is real but it is not the gap the sales page implied.

Then there is the part nobody publishes. No offshore provider discloses what it pays its assistants. The Philippine government benchmark sits near $2 an hour locally, and advertised agency prices for the same category of work run several times that.

Some of that gap is a genuine premium for foreign-client work in US hours. The rest is margin. That split is undisclosed and unmeasured, and treating our arithmetic as anything more than inference would be dishonest.

If you are comparing an agency quote against a domestic hire, load both sides. The BLS median is a wage, not a cost. A domestic employee at $23.66 costs your firm more once payroll taxes and unemployment insurance land on top.

Month three is when you fire your accounting assistant. Month six is when it works

Month three is the trough and month six is the turn. This is the most useful pattern in the whole research base, and it is entirely absent from the pages selling the service.

One owner describes reaching the edge: "at 3 months I was close to firing him". The same person describes the same hire at six months as a pleasure to work with, going beyond what was asked.

The first ninety days are net negative on purpose. You stop billable work to document procedures, record training, and review output line by line. That is the investment, and there is no version of this that skips it.

Not every relationship turns. One owner reports offshore contractors taking far longer to onboard than domestic staff, and adds: "Even after a year, I'm still training him on things". Failed relationships usually break before month six on repeated mistakes, missed deadlines, or a candidate who quietly disappears mid-cleanup.

There is no sector research on this. The onboarding statistics circulating on hiring blogs could not be traced to a primary study, so we are not repeating them. The one usable idea comes from research on complex implementation roles, where self-sufficiency took three to twelve months, and productive meant work that no longer needed checking by someone else.

If you cannot fund three to six months of review time on top of the wage, you are not ready to hire yet.

View-only bank access breaks once your accounting practice has twenty-five client logins

View-only access is the right control and it does not scale cleanly across a book of clients. One owner found the limit fast. The bank offered third-party view-only access, but required a separate login per account, so "I'd be juggling 25+ sets of credentials".

The rule underneath is not negotiable. A remote assistant should never hold credentials that can move money. They can still pull statements, categorize feeds and complete reconciliations without that ability.

The practical bottleneck is the second factor. Bank codes arrive on a client's phone, and the workflow stops until someone forwards them. Firms solve it with an enterprise password manager that stores the one-time-password seed, so the assistant generates their own codes.

Shared plain credentials are the failure mode everyone recognizes. As one owner puts it, "there would never be shared credentials and there would be 2FA implemented". Some firms go further and ship a locked-down machine with disabled ports and browsing restricted to the tools the work needs.

If you have five clients, a shared vault with each client is workable. If you have fifty, per-client access control is a system you have to design, not a setting you can switch on.

Two obligations sit behind all of this. Your firm is a financial institution under the Gramm-Leach-Bliley Act, so the FTC Safeguards Rule applies at any size. And the AICPA Code requires you to satisfy yourself that a provider has the skills and safeguards before work starts.

Your accounting firm's E and O policy may not cover a remote contractor

Your coverage may not reach the contractor, and this is the hardest fact in the research. One firm had a remote contractor's machine compromised and fraudulent vendor payments made. The firm held errors and omissions cover, fraud cover and cyber cover.

The carriers pointed at each other over the contractor's status, the contractor's location and the client-owned system that was used. The owner's summary: "A year later and no one has paid yet".

A second firm describes a different route to the same place. An overseas team member "clicked on something that infiltrated our network", which cost an outside remediation firm and a week of downtime during filing season.

The insurer guidance is more specific than most owners expect. CNA, which underwrites the AICPA professional liability program, notes that firm coverage generally extends to subcontractor work subject to policy terms. It still recommends the provider carry its own.

CNA also recommends contracting for minimum data security requirements, indemnification for a security incident, and restrictions on further subcontracting. Choice of law and forum matters most when the provider is offshore.

CAMICO, a CPA-owned insurer, goes further on disclosure than the rules require, encouraging written client consent for all offshore engagements as defensive documentation.

Take one action from this section. Call your broker before the engagement starts and get the answer in writing. If the answer is conditional, you have found the condition while you can still meet it.

Payroll does not exempt your accounting firm from the disclosure rules

Payroll status does not solve disclosure. One owner states the mechanism plainly: if the person sits outside the country, "you are still disclosing taxpayer information outside the United States".

Section 7216 of the Internal Revenue Code makes it a criminal offense for a preparer to disclose tax return information without consent, and disclosure outside the US requires consent in a prescribed format for 1040 clients. There is an additional rule on Social Security numbers, which must be masked unless the disclosure runs through a recognized data protection safeguard.

The rule almost nobody quotes sits in the AICPA Code. Interpretation 1.150.040 requires you to inform the client, preferably in writing, before confidential information reaches a third-party service provider. If the client objects, you either drop the provider or decline the engagement. The client holds a veto and it applies onshore as well as offshore.

Owners handle this four different ways, and the disagreement is genuine. Some run a separate consent for offshore preparation and price domestic-only work higher. Others disclose in the engagement letter with a right to object.

Some believe no notification is owed at all, reasoning that clients buy the output. And one owner describes being instructed not to discuss the arrangement: "we were instructed not to discuss the offshore teams with clients".

Firms have lost clients over this. In the research, clients terminated after discovering offshore handling they had not agreed to.

One thing the Code does not settle. Whether a directly hired, firm-supervised assistant is a third-party service provider at all is unresolved. The interpretations were written for outsourcing providers. Hiring through an agency puts those interpretations plainly in play.

Bookkeeping yes, controllership no. Where accounting firm owners draw the line

Owners draw the line at judgment, and they draw it consistently. One puts it in a sentence: "I would never outsource controllership or CFO work".

What goes out is transactional and repeatable. Bank feeds and categorization. Reconciliation preparation. Invoice entry with payments staged for your approval. Payroll journal entries. Raw data entry into return software.

What stays in is everything carrying professional judgment. Final review and the signature. Advisory and forecasting. Complex tax resolution. Anything where you assume the liability, which is most of it.

One firm outsources 60 to 80 percent of individual returns, including the larger ones, while keeping all corporate work and reviewed financials in house. That split is a useful model even if your percentages differ.

Client contact is the boundary owners guard hardest. Most route communication through a ticketing system, with the assistant drafting and the owner sending. The fear is specific and reasonable, that an assistant answers a question that was actually tax advice.

Two limits are not yours to move. Anyone preparing a substantial portion of a return is a tax return preparer under the regulations, needs a PTIN, and carries preparer penalty exposure, whether or not their name appears. And attest work is generally walled off to licensed firms, with the specifics varying by state board.

If you delegate reconciliations without a fixed review step, expect the failure the research describes. Differences get plugged to retained earnings and the balance sheet stops meaning anything.

Worker classification when your bookkeeping virtual assistant is overseas

Classification is cleaner offshore than most owners expect, and messier domestically. Take the two cases separately.

For a person outside the US, the sourcing rule decides it. Compensation for personal services is sourced where the services are physically performed. Not where the contract was signed, not where payment came from, not the worker's citizenship.

So a non-US person working entirely abroad earns foreign-source income. No 1099-NEC. No thirty percent withholding. No 1042-S. What you do need is a signed Form W-8BEN before the first payment, kept in your records rather than filed, valid for three calendar years.

Three traps sit around that. No W-8 on file and the IRS may treat the payee as a US person, which pulls the payment into 1099 reporting and can trigger backup withholding. A US citizen or green card holder living abroad remains a US person, so a W-9 applies and the foreign address changes nothing.