How do you start a virtual assistant business in 2027?

The generalist VA versus the vertical specialist practice

There are only two real ways to start a virtual assistant business, and they diverge on day one. The first is the default path: register on a marketplace, describe yourself as a virtual assistant who can "help with anything," price by the hour somewhere between $15 and $35, and accept whoever says yes. The second is the specialist path: choose an industry, choose a single job title inside that industry, define a scope, and sell it as a monthly retainer with a fixed price and an outcome attached.

The generalist option is genuinely faster to start. You can have a profile up in an afternoon and a first client inside two weeks. It requires no positioning decision, no vertical research, and no confidence in your own pricing. That speed is exactly why most people take it, and it is also why most VA businesses stall out somewhere between $45,000 and $70,000 of annual revenue and stay there.

The structural problem with the generalist option is that it has three simultaneous competitors and loses to all of them on the only axis it competes on. It loses to offshore generalists on price, because a competent English-fluent VA in the Philippines or Latin America bills $4–$15 an hour and you cannot. It loses to AI agents on speed and cost, because inbox triage, calendar scheduling, meeting summarization, basic web research, transcription, and simple data entry are precisely the task categories that 2027-era agentic tools handle without a human in the loop. And it loses to specialist firms on trust, because a buyer comparing "a VA who does anything" against "a transaction coordinator for residential real estate teams" has no way to evaluate the first except by price.

The specialist option is slower to start — you spend two to four weeks on positioning before your first outreach — and it carries a real cost: you must say no to off-niche clients while you are hungry, which feels irrational and is the single hardest discipline in the first six months. But it inverts every one of those three competitive losses. Against offshore generalists you compete on vertical knowledge, time-zone alignment, and ownership of an outcome rather than a task. Against AI you compete on judgment, exception handling, accountability, and the fact that you operate the AI stack rather than being displaced by it. Against other agencies you compete on niche depth, which is available to anyone but chosen by almost no one.

The economic difference between the two options is not marginal. A generalist at $30 an hour, fully booked at 30 billable hours a week, tops out around $46,000 of revenue with zero paid time for selling, onboarding, or admin. A specialist running six clients on $2,400 monthly retainers is at $172,800 with a lower total workload, because the retainer decouples income from keystrokes. Every hour AI saves the generalist is an hour they cannot bill. Every hour AI saves the specialist is margin they keep.

There is a third variant worth naming because people conflate it with the second: the productized package. Instead of selling a role by the month, you sell a countable deliverable at a fixed price — four fully produced podcast episodes for $2,400, or $450 per closed real estate transaction. This is a subtype of the specialist path rather than a separate strategy, and it works best where the output is discrete and countable. It is the easiest model to hand to a contractor later, because the deliverable is already the specification.

How to decide which model to build

The decision comes down to four inputs: what you already know, how AI-resistant the role is, whether your target buyer has decision authority, and whether you want a job or a company. Work them in that order, because the later questions are meaningless if you fail the earlier ones.

Start with prior exposure. Vertical credibility is the thing you cannot fake in a discovery call, and it is what lets you charge $2,400 instead of $1,200 for the same hours. If you spent six years as an in-house executive assistant, your vertical is fractional executives and your role is EA. If you ran the back office at a real estate team, your vertical is residential real estate and your role is transaction coordinator. If you have no relevant background at all, pick the vertical you are most willing to read about obsessively for three years, and expect the ramp to positioning credibility to take six to nine months rather than zero.

Then test the role for AI resistance. Ask whether the work involves managing other humans, handling money-adjacent processes, navigating regulated workflows, or owning a relationship-heavy function. Roles that clear at least two of those tests — transaction coordination, community management, embedded e-commerce operations, EA support for a multi-client fractional — are durable. Roles that clear none — transcription, data entry, basic scheduling, generic research — are the ones being absorbed. This test matters more than market size, because a large market that AI is eating is worse than a small one it cannot.

Third, check buyer authority. The best VA clients can say yes on a thirty-minute call with no procurement, no committee, and no budget cycle. Solo founders doing $300K–$1.5M, fractional executives, and owner-operators of $1M–$5M businesses all clear that bar. Seed-stage startups mostly clear it. Enterprises never do, and you should not target them until you have a real agency with a security posture to match.

Finally, decide what you are actually building. This determines everything downstream — hiring, documentation discipline, whether you keep client relationships in your own hands, and whether the business has resale value at all. Solo produces a ceiling around $160,000–$200,000 with near-total margin and no enterprise value. Agency produces $300,000–$1.4M with 22–38% net margin and a sellable asset. Both are legitimate; picking accidentally is not.

One more filter is worth applying before you commit: count the businesses in your intended ICP band. The United States has roughly 33 million small businesses, but the overwhelming majority are non-employer sole proprietorships with no budget for delegated help. The band that actually buys — firms doing $300K–$3M with one to ten people, enough cash flow to pay and enough chaos to need it — plausibly numbers between six and nine million. Add the creator economy and the fast-growing fractional-executive cohort and the pool is comfortably in the millions. You need five to fifteen of them. If your intended niche cannot plausibly contain a few hundred thousand potential buyers, it is too narrow; if you cannot name the buyer's job title, it is too wide.

The numbers behind each option

Launch capital is trivially low, which is both the appeal and the reason competition is dense. Budget $1,500–$6,000 for one-time setup: $0–$500 for LLC formation depending on state, plus $50–$800 annually for a registered agent and state fees; $400–$1,500 for an attorney to draft or review your client services agreement and contractor agreement; $400–$900 a year for professional liability and general liability coverage; $200–$800 for a domain and a one-page site, which is genuinely sufficient at launch; $0–$600 for minimal branding; and $150–$500 for initial tooling.

Recurring costs for a solo operator land between $150 and $450 a month. A business-tier password manager runs $4–$8 per user and is non-negotiable, because it is how you hold client credentials without holding raw passwords. Project management — ClickUp, Asana, Notion, or Trello — costs $0–$25. Communication and scheduling adds $0–$40. AI tooling is now $20–$120 and belongs in the core cost line, not the optional one. Accounting software is $15–$40, async video is $0–$15, and security tooling adds $5–$15.

Unit economics diverge sharply by model. Take a Core-tier client at $2,200 a month. Solo, your cash cost is software allocation of roughly $40–$70, so contribution margin is above 95% and your labor is the owner's draw. Route that same client to a contractor VA costing $1,000–$1,800 a month full-time-equivalent across one or two clients, and direct cost per client becomes $700–$1,400 — gross margin compresses to 35–55%, and after your management time and overhead a well-run VA agency nets 22–38%.

Pricing structure is where most of the revenue difference lives. Three tiers cover nearly every buyer. Lite runs $900–$1,400 a month for roughly 25–35 hours: one defined role, part-time coverage, async communication, a weekly check-in — right for early-stage solo founders and creators. Core runs $1,800–$2,600 for roughly 55–70 hours with deeper ownership, faster response windows, and a standing weekly call; most clients land here and it should be the tier you actively sell. Embedded runs $3,200–$5,500 for near-full-time dedicated support from a named person who functions as part of the team, sometimes spanning two roles — appropriate for funded startups and established owner-operators who would otherwise make a hire.

Always charge an onboarding fee of $350–$1,500. It funds discovery, SOP intake, credential setup, and the first two weeks of heavy training that are otherwise unbillable, and it filters out buyers who were never going to commit. Build a 4–9% annual rate-review clause into every agreement. If you ever bill hourly — and the only defensible uses are unpredictable overflow work or a short paid trial — set it at $55–$85 for specialized work, because a low hourly rate anchors every negotiation you will ever have with that client.

Willingness to pay by segment is reasonably consistent. Solo founders and bootstrapped operators pay $1,200–$2,800. Fractional executives pay $1,800–$3,500 and refer each other relentlessly, which makes them the highest-leverage first segment. Lean funded startups pay $2,500–$5,000 but carry churn risk when they hire in-house. Established owner-operators — real estate teams, clinics, law firms, home-services companies — pay $1,500–$4,000 and are extremely sticky once embedded. Creators pay $1,000–$3,500 and are excellent for portfolio and referrals but financially volatile.

The revenue trajectory follows from the model, not the industry. Year one on the specialist path realistically produces 4–9 retainer clients and $55,000–$120,000: months one through three go to positioning, contract, site, tooling, and one to three clients often sourced from your existing network, totaling $3,000–$15,000; months four through six reach $6,000–$12,000 monthly as delivery tightens and referrals start; months seven through twelve reach $8,000–$16,000 monthly with raised rates for new clients. Year two lands $120,000–$240,000 and is where you commit to solo or agency. Year three is $160,000–$200,000 optimized solo, or $240,000–$480,000 on the agency path with three to five contractors and twelve to twenty clients. Year four reaches $400,000–$800,000 with six to twelve contractors. Year five reaches $600,000–$1.4M and forces a choice: keep scaling, sell at roughly 2.2x–3.6x SDE, or evolve into a fractional-COO practice with fewer, larger clients.

On exit economics, be clear-eyed. A solo VA practice produces almost no enterprise value, because you cannot sell a job. An owner-independent agency in the $400K–$1.5M range with documented systems, low churn, and a stable team sells to roll-ups at roughly 2.5x–3.6x SDE, to competitor agencies at 2.0x–3.0x, and to individual operators at 1.8x–2.6x. Typical structure is 50–70% cash at close, 20–30% seller note over 24–36 months, and 10–20% earn-out tied to client retention, with a two-to-four-year non-compete. A $700,000-revenue specialist agency at 30% SDE is roughly $210,000 of SDE and about a $630,000 sale. What crushes the multiple: founder does all delivery, hourly billing, client concentration, no documentation, contractor relationships that leave with the contractor.

Marketing spend stays small. A serious year-one budget is $1,500–$4,000, covering a site, a content tool or freelance writer, podcast equipment, and community memberships — almost no paid advertising, because VA buyers buy on trust and trust is not purchasable through CPCs.

Sequencing the launch and the first twelve months

The order of operations matters more than the individual steps, because several of them are irreversible in practice. Pricing is the clearest example: a client acquired at $800 a month for Core-tier scope is a client you will resent inside ninety days and cannot re-price without a painful conversation. Set the rate before you take the first call.

Weeks one through three are positioning and paper. Lock the vertical-and-role pair and write it as a job title — "Executive Assistant for fractional CFOs and COOs," "Transaction Coordinator for residential real estate teams," "E-commerce Operations Assistant for Shopify DTC brands." If it does not read like something a company would post, it is not yet a position. Form the LLC. Get the client services agreement drafted or reviewed, which must define scope and explicit out-of-scope, fee and billing date and auto-renewal, the hours band and what happens on overage, response-time expectations, confidentiality and data handling, IP assignment to the client, limitation of liability, 14–30 day termination either way, and non-solicitation if you will ever use contractors. Bind insurance. Stand up the password manager, project management workspace, accounting software, and AI tooling.

Weeks four through eight are your first clients and your first SOPs simultaneously. Work your existing network first — the highest-converting year-one clients almost always come from people who already know your work. Run structured onboarding on every one: a discovery call to map pain and define role scope, a brain-dump session where the client walks you through their world, credential access through the password manager, and an initial SOP library where every recurring task is documented as a repeatable process before you have done it twenty times. Set communication norms explicitly — where you talk, response-time expectations, the standing weekly call, how work gets requested and prioritized. Onboarding quality is the difference between a three-year client and a three-month one.

Months three through twelve are cadence and channel. The daily rhythm is a morning review of each client's channels and board, execution batched by client to limit context-switching, AI-assisted first passes followed by your human review, and an end-of-day async note to every active client so they always feel covered. Weekly means a standing call or async video per Core and Embedded client covering what shipped, what is next, what you need, and what you are flagging. Monthly means a written recap of outcomes and hours-used-versus-band, plus an internal review of revenue, contractor cost, margin per client, and any account whose scope has crept past its tier. Scope creep is the silent margin killer and must be caught on a monthly cycle, not an annual one. Quarterly means a strategic check-in with Embedded clients that positions you as an operations partner rather than a task-doer — this is the retention and upsell engine.

Channel sequencing follows a reliable order. Referrals are the dominant channel for specialists and should be engineered rather than hoped for: ask explicitly at the ninety-day mark, make the ask easy, and consider a one-month-at-half-price or flat $200–$500 thank-you. Mature specialist practices see 45–65% of new clients arrive this way. Niche content is second — not "what is a virtual assistant" but the narrow queries your exact buyer types, like a transaction-coordinator checklist for real estate teams or how a fractional CFO should structure EA support. Thirty to sixty deep pieces compound into 1,000–8,000 relevant monthly visits after twelve to twenty-four months, converting at 1–3% to a discovery call. Podcast guesting is third and unusually high-leverage: two appearances a quarter on shows your ICP already listens to yields a handful of qualified leads each plus referral relationships. Community presence and, for B2B niches, a focused LinkedIn cadence of three to five posts a week round it out. Partnerships with bookkeepers, fractional CMOs, web designers, and CRM consultants who serve the same buyer but do not compete can drive 20–35% of pipeline. Marketplaces are a minor supplementary channel for a sharply positioned specialist and a race to the bottom for anyone else. Build three to four independent channels before you consider the business stable — 80% of leads from one podcast or one partner is a cliff, not a channel.

The hiring trigger is specific: six to eight stable clients plus a small waitlist. Earlier than that and you are handing contractors chaos instead of a system; later and you burn out while damaging the relationships you built. Source your first contractor through your own network of specialist VAs at capacity, then niche communities, then curated platforms. Offshore full-time-equivalents in the Philippines and Latin America run $1,000–$2,200 a month; US-based contractors run $20–$40 an hour or $2,500–$5,000 monthly FTE. Budget sixty to ninety days of training even with excellent SOPs, have the contractor shadow you on a client for two to four weeks before owning anything, and keep the client relationship in your own hands during the early stages so a departure is an inconvenience rather than a crisis. Around $250,000–$400,000 of revenue you need a lead VA or operations manager whose job is partly reviewing other VAs — that hire is what stops you being the bottleneck for quality and is the gate to scaling past $400,000.

Two operational disciplines determine whether any of this holds. The first is documentation: every recurring task becomes an SOP from day one, because a business living entirely in the founder's head cannot scale, cannot be sold, and cannot survive a vacation. The second is security posture, which is genuinely part of the product — you hold credentials, financial data, and sometimes regulated information. Business-tier password manager, two-factor everywhere, VPN, encrypted storage, a written data-handling policy, and immediate deprovisioning when a contractor or client relationship ends. One breach can end a VA business, because clients hand you the keys to their operation.

Know the regulatory edges of your specific niche before you sell into it. A general VA business needs no special license beyond basic local registration, but verticals carry lines you must not cross: real estate transaction coordination has a boundary between coordination and licensed real estate activity; bookkeeping-adjacent operations work requires your contract to state plainly that you are not a bookkeeper or accountant; touching healthcare data may make you a HIPAA business associate requiring a BAA. Worker classification is the other live risk — contractor agreements need real contractor-style working relationships behind them, and an employment attorney should review your setup as you scale.

Where AI sits in the 2027 positioning

The framing you choose for AI in client conversations is a pricing decision disguised as a messaging decision. The wrong frame is "I use AI so I'm cheaper," which positions you as a commodity layer and invites the buyer to reason their way to cutting you out. The right frame is that you operate a stack of AI tools on their behalf, supervise the output, catch the errors, and own the result — they get the speed of AI with the accountability of a person.

This is not spin; it maps to how the work actually decomposes. AI produces the first draft of an email, the meeting summary, the research brief, the structured data pull, the support-ticket response. You supply the judgment about whether the draft reads correctly for this particular recipient, whether the summary missed the political subtext, whether the data is being used to answer the right question, and whether the exception in front of you needs escalation. You are also the person the client texts at nine at night when a vendor payment fails, and you are the entity that can be held accountable when something goes wrong. That accountability is a substantial part of what the retainer buys.

The pricing consequence is direct and is the reason retainers beat hourly in this specific era. Under hourly billing, every efficiency AI delivers is revenue you lose — you get faster and earn less, which is a structurally insane incentive. Under a monthly retainer for a defined role, the same efficiency is margin you keep. This is why the 2027 answer on pricing is not a preference but close to a requirement: the pricing model determines whether AI adoption helps or hurts you.

The practical delivery stack varies by role but has a consistent 2027 core: a paid general assistant tier for drafting, summarizing, research, and increasingly agentic multi-step execution under supervision; AI-native scheduling and inbox-triage tools that propose drafts and times for your approval; transcription and meeting-notes tooling; role-specific layers like clip generation and show-notes for creator work, or CRM lead-scoring and follow-up drafting for sales-ops work; and whatever vertical software your niche runs on. Competence with these tools is itself billable, because most of your clients do not have time to evaluate, configure, and supervise them.

The strategic read on AI as competitor is that it is not something you beat, it is something you absorb. It is shrinking the segment you were never going to win — generic hourly execution — while leaving intact and growing the segment where judgment, relationship, compliance, and coordination complexity dominate. Treat it as the reason to specialize and move upmarket on a continuing basis, not as a reason to panic. As agent capability rises each year, the honest expectation is that the floor of what counts as "specialized" rises with it, and your positioning needs a deliberate review annually rather than a one-time decision at launch.

The mistakes that end year one

The failure patterns are remarkably consistent, and nearly all of them are positioning or pricing errors rather than delivery errors. Positioning as a generalist rather than a vertical-and-role specialist is the root cause of most of the rest. Pricing by the hour at marketplace rates instead of monthly for a defined role caps income and inverts your incentives. Starting low "to get experience" creates clients you cannot re-price. Skipping the onboarding fee out of desperation removes both your unbillable-time coverage and your best qualification filter.

Taking clients outside the chosen niche destroys the systematization that makes contractors and eventual sale possible, and it dilutes the credibility that lets you charge specialist rates — one off-niche client is a rounding error, four of them means you are a generalist again. Not documenting SOPs from day one means the business lives in your head, which caps it permanently. Operating without a real services agreement and limitation-of-liability clause turns one bad month into a legal exposure.

Letting a single client reach 40–50% of revenue makes their departure existential; hold every account under roughly 20–25% and keep a small waitlist as a buffer. Hiring contractors before your SOPs and sales engine can support them produces quality collapse and client churn simultaneously. Allowing scope creep to run unexamined for months quietly converts a profitable retainer into unpaid labor. And relying entirely on marketplace bidding for leads puts you in direct price competition with exactly the two competitors you cannot beat on price.

The last one worth naming is a self-knowledge failure rather than a business one. VA work is service work, and service work carries an emotional load — clients have urgent problems, bad days, and occasionally unreasonable expectations, and you absorb all of it. Solo operators who never set boundaries collapse regardless of how good the financials look. Build response-time windows into the contract, take real time off, default to async communication, and be honest with yourself early about whether you want a high-autonomy job or a company you will eventually manage rather than do. Most people who ask how to start a virtual assistant business actually want the first, and the first is a genuinely good outcome — it is only a failure when you build it by accident while telling yourself it was the second.

Related questions

What niche should I pick if I have no relevant experience?

Pick the vertical whose vocabulary and software you are willing to study obsessively, then choose the most judgment-heavy role in it. Expect six to nine months before you carry real credibility. Community presence and niche content close that gap faster than certifications do.

Can I start a VA business while employed full-time?

Yes, and it is the lower-risk path. Two Lite-tier clients at $900–$1,400 fit around a job at roughly ten to fifteen hours a week. Check your employment agreement for moonlighting and non-compete terms first, and never serve your employer's clients.

Should I use Upwork at all?

As a minor supplementary channel with a sharp specialist profile and premium pricing, yes — it provides social proof and occasional inbound. As your primary channel it is a bidding war against offshore labor and AI, and it will hold your rates down indefinitely.

How many clients can one solo VA realistically hold?

Four to eight retainer clients depending on tier mix. Six Core-tier accounts is roughly a full load at sustainable hours. Beyond eight demanding clients you are choosing between capping out and starting the multi-year transition to an agency model.

When do I raise rates on existing clients?

At contract renewal using a rate-review clause you inserted at signing, typically 4–9% annually. Raise outside that window only when scope has genuinely expanded, and frame it as a tier change with new scope rather than a price increase on the same work.

FAQ

Do I need an LLC to start a virtual assistant business?

Not legally in most jurisdictions — you can operate as a sole proprietor — but form one anyway. It costs $0–$500 plus $50–$800 a year in state fees and registered agent costs, separates personal and business liability, and reads as professional to buyers. Consider an S-corp election once net profit is consistently above roughly $60,000–$80,000, with an accountant's input on the threshold for your situation.

How much can I realistically earn in the first year?

$55,000–$120,000 on the specialist path with four to nine retainer clients, assuming you priced correctly and had some relevant background. Expect $3,000–$15,000 total across the first three months, then $6,000–$12,000 monthly by month six. A generic hourly practice follows a different and flatter curve, typically plateauing at $45,000–$70,000.

What if a client asks me to do work outside my defined scope?

Handle it explicitly and quickly. Small one-off requests are goodwill and cost nothing. A pattern of out-of-scope requests is a tier conversation: show the hours-versus-band data from your monthly recap, name the new responsibilities, and quote the higher tier. Doing this at month two is a routine business conversation; doing it at month ten after silent absorption feels to the client like a surprise price hike.

Is offshore competition a real threat to a US-based specialist?

Not at the specialist tier. Offshore generalists compete on hourly price in the $4–$15 band, which is a market you are not in. Where you overlap is generic execution work, and that segment is being absorbed by AI regardless of who staffs it. Specialists compete on vertical knowledge, time-zone alignment, accountability, and owning an outcome — none of which are price-comparable.

What should be in my client services agreement?

Scope and explicit out-of-scope, fee with billing date and auto-renewal, the hours band and overage handling, response-time and availability expectations, confidentiality and data handling, IP assignment of work product to the client, a limitation-of-liability cap, termination on 14–30 days notice either direction, and non-solicitation covering your contractors. Pay $400–$1,500 for an attorney to draft or review it — it is the cheapest risk control you will buy.

Should I hire contractors or employees?

Contractors early, for flexibility and better economics at small scale. Every contractor signs an agreement with confidentiality, IP assignment, non-solicitation of your clients, and clear classification language, and the working relationship must genuinely match contractor status. Worker-classification rules vary by jurisdiction and penalties are real, so get an employment attorney's review before you are managing several people.

Sources

• U.S. Small Business Administration — Office of Advocacy small business data
• IRS — Independent contractor (self-employed) or employee?
• U.S. Bureau of Labor Statistics — Occupational Outlook Handbook, Secretaries and Administrative Assistants
• SBA — Choose a business structure
• U.S. Department of Health and Human Services — HIPAA business associates
• Federal Trade Commission — Business guidance for small businesses
• NIST Small Business Cybersecurity Corner
• IRS — S corporation election and requirements
• U.S. Department of Labor — Misclassification of employees as independent contractors

Related on PULSE

• How do you price a productized service retainer instead of billing hourly?
• What does a fractional executive actually delegate to an assistant?
• How do you build an SOP library that a contractor can actually run?
• When should a solo service business make its first contractor hire?
• How do RevOps teams decide what to automate versus what to staff?
• What is a realistic valuation multiple for a small service agency?