State of Virtual Assistant Services 2026 (first-party report)

State of virtual assistant services 2026 is a compilation of published third-party data on a market that three separate research houses size at $6.5 billion, $19.5 billion and $29.1 billion, all for the same year. This report sets out each figure with its definition and source, then covers adoption, workload mix, pay and pricing across four English-speaking markets.

Executive summary

Ten findings, each sourced in the chapters below.

1. Market size estimates for 2026 range from $6.5 billion to $29.1 billion depending on where the boundary is drawn between VA agencies, human virtual staffing and adjacent staffing categories.
2. Growth estimates are far more consistent than size estimates, clustering between 23.4 percent and 25.9 percent compound annual growth.
3. Because that growth rate is applied over nine years, 2035 forecasts diverge dramatically, from $43.4 billion to $231 billion.
4. About 69 percent of US businesses use some form of virtual assistant support, on the widest definition.
5. Small and medium businesses make up 44 percent of clients, which explains why the market prices in 20 hour and 40 hour weekly plans.
6. Administrative assistants are 31.5 percent of the market, the largest single category.
7. Administrative and marketing work together account for more than 62 percent of assistant workloads.
8. US pay benchmarks disagree by about 40 percent: PayScale reports $19.45 an hour, Indeed $27.15.
9. Pricing separates into four tiers with an order of magnitude between the extremes: marketplace around $13 an hour, managed offshore $7 to $15, premium onshore about $65, and AI receptionists at $25 to $300 a month flat.
10. Compliance load varies sharply by market, and it is now a staffing design question rather than a legal afterthought.

Methodology, and what is not in this edition

This edition contains third-party published data only.

What was included. Figures published in 2025 or 2026 by named market research firms, salary data platforms, government statistical agencies and provider rate cards. Each figure appears as its source stated it, in the currency the source used, with the year attached.

What was excluded. Any figure without a named source. Any figure that required averaging two incompatible estimates. Any aggregate reported without a definition of what was being counted. Any provider claim about its own results that could not be checked against a published page.

What was not done. No survey was conducted for this edition. No client data was collected. No operational metrics were modeled. Where a number does not exist in the public record, this report says so rather than estimating it.

On first-party data. AssistBPO’s own first-party dataset is not in this edition. It arrives in the next one. To be plain about it: AssistBPO has not yet completed a documented collection of its own operational data to a standard that would survive citation, so publishing anything of that kind today would be an estimate wearing a research jacket. The next edition is planned to cover answer times and answer rates across staffed desks, ticket handling measures, onboarding time to full productivity, and coverage patterns by market and time zone, with the collection window, sample definition and exclusions all published alongside the numbers. Until that is done, this report stays with other people’s data and credits them for it.

Review cadence. Quarterly. The updated date at the top reflects the last review. Published rate cards change without notice, so verify any figure you intend to rely on.

Chapter 1. Market size and growth

Three sources, three boundaries, three answers.

Two observations.

First, the disagreement is definitional rather than empirical. A firm counting agency revenue and a firm counting all human virtual staffing will produce different totals from the same underlying activity. Neither is wrong. What is wrong is quoting the largest number without the definition that produced it, which happens constantly in vendor marketing.

Second, the growth band is the durable finding. Between 23 and 26 percent a year, sustained, from three independent starting points, is a more meaningful signal than any single headline total. Applied over nine years that band is also why the 2035 forecasts differ by a factor of five. Treat 2035 numbers as directional and the growth band as the finding.

Chapter 2. Who is buying virtual assistant services?

The 69 percent adoption figure needs a caveat attached every time it is used. It is a broad-definition number, covering everything from a handful of task hours a month to a full-time dedicated assistant. It should not be read as 69 percent of US businesses employing a permanent remote team. What it does establish is that buying assistant hours is now an ordinary purchase rather than an experiment.

The 44 percent SMB share is the more useful number for anyone designing a service. It explains why nearly every published plan in this market is built around 20 hours or 40 hours a week, why month-to-month terms are the norm, and why replacement policy is a headline feature rather than a contract footnote. Small businesses cannot absorb a bad hire, so they buy supervision as much as they buy hours.

The 475 percent growth figure since 2020 is the clearest evidence that the pandemic shift to remote work did not reverse for this category. It normalized the idea that an administrative colleague need not sit in the building.

Chapter 3. What the work is

Nearly two thirds of what assistants do is inbox, calendar, documents, data, social media and content. This concentration has three consequences.

Pricing pressure sits in the middle. The administrative band is the most competitive part of the market because it is the largest and the least differentiated. Specialist work, including bookkeeping support, insurance verification, prior authorization follow-up and technical support, prices above it and has fewer providers.

Phones are a gap. Most providers in this market are typing-only. Live call handling requires telephony, scripting, recording compliance and shift coverage, which is a different operational build from an inbox desk.

Role-based buying is replacing task-based buying. The published plans that dominate the market are structured around a person who owns a function, not a pool that completes tickets. That is consistent with a market where SMBs are the majority of clients and continuity matters more than elasticity.

Chapter 4. What does a virtual assistant cost by market?

United States

The PayScale and Indeed gap of nearly $8 an hour on the same job title deserves attention rather than an average. The two platforms sample differently, with Indeed weighted toward employed full-time postings. A buyer should see both figures and understand that “US-based assistant” is a range, not a price.

One recurring figure from the 2026 cost guides belongs here: Zedtreeo and Catalyst Outsourcing estimate that losing a freelance assistant costs two to three months of their pay in lost productivity while a replacement is recruited and trained. Over a two-year horizon a single replacement absorbs most of the rate gap between a marketplace hire and a managed one, which is the strongest available argument for the managed model at the price it commands.

United Kingdom

Two UK-specific factors distort headline comparisons. Most agencies are VAT registered, so a client who cannot reclaim VAT pays a 20 percent uplift on the quoted rate, turning £28 an hour into £33.60. And UK buyers use different vocabulary: virtual PA, diary management, telephone answering, call handling, minute taking. A provider that advertises only in American terms will be invisible to half its UK market.

Australia

Australia has the widest gap between offshore and local cost of the four markets, and the buying criterion is time zone rather than price. A provider that cannot genuinely staff Australian Eastern hours is selling a partial overlap, and buyers here have learned to ask the question directly.

Canada

Canadian buyers ask two questions the other markets do not. Whether French coverage is genuine fluency or a capability listed on a website, and how the provider handles CASL, which is the strictest anti-spam regime of the four markets.

Chapter 5. Can a virtual assistant answer your phones?

Phone answering is priced on a different model from assistant hours, and the contrast is instructive.

Source for the first three rows: OnceHub’s September 2026 roundup.

The per-minute model is the loudest complaint in this category. A plan that looks affordable at 50 minutes becomes a four-figure bill in a busy month, and the buyer has no control over call length. That structural problem is what created room for two alternatives: AI receptionists at a flat monthly price, and dedicated assistants who answer the phone as part of a staffing plan rather than per minute.

The AI band at $25 to $300 a month is roughly two orders of magnitude below a staffed desk, which means price is not the comparison that matters. The comparison that matters is what happens on a call that does not follow the script: a distressed patient, a caller with an accent the model mishandles, a complaint that needs judgment. Most businesses in 2026 are landing on a split, with automation handling simple status calls and people handling everything with consequences.

Chapter 6. The five models buyers choose between

The pattern across the published data is that the market has stopped competing on hourly rate alone and started competing on continuity: who supervises the assistant, who covers absence, and how quickly a replacement is in place. That is a direct consequence of the SMB-heavy client base.

Chapter 7. The compliance load by market

Compliance is now a staffing design decision, because the rules determine what a remote desk may do and how it must be logged.

The practical consequence is that a provider serving all four markets needs four different outbound playbooks, four sets of calling windows and four consent models, enforced by systems rather than by memory. That is a build cost, and it is one reason the managed tier prices above the marketplace tier.

Chapter 8. What to watch into 2027

Four things worth tracking, stated as questions rather than predictions.

Does the growth band hold? Three sources agree on 23 to 26 percent for 2026. A meaningful deviation in any of them next year would be the first real signal of the market maturing.

Does the AI price floor move the human market? AI receptionists at $25 to $300 a month flat have not yet visibly compressed staffed desk pricing. The interesting measure is not whether AI replaces assistants but whether it changes what a staffed hour is expected to cover.

Does specialist work outgrow administrative work? Administrative work is 31.5 percent of the market today. If regulated and technical specialisms grow faster, the pricing shape of the whole category changes.

Does the definitional gap close? As long as credible sources report $6.5 billion and $29.1 billion for the same year, buyers will keep encountering contradictory claims. A shared boundary would make every figure in this report more useful.

Limitations

This report has four honest limits, stated so readers can weigh it properly.

1. No primary research. Every figure is someone else’s. Nothing here was collected for this report.
2. Definitional inconsistency. The three market size estimates are not directly comparable, and no attempt was made to reconcile them.
3. Rate cards move. Published provider pricing reflects pages as at September 2026 and changes without notice.
4. Sample bias in pay data. PayScale and Indeed rely on self-reported and posting-derived data respectively, and both skew in ways neither platform fully discloses.

How to cite this report

Cite as: AssistBPO, State of Virtual Assistant Services 2026, published 19 September 2026. Where you are quoting a specific figure, cite the underlying source named beside it as well, because the number belongs to that organization. Every source used appears in the list below.

AssistBPO is the virtual assistant and remote staffing division of SS Support Network LLC, a business process outsourcing company registered in Vancouver, Washington, with a second office in Pakistan. It staffs dedicated, employed assistants, receptionists, customer support and back-office teams working US, UK, Canadian and Australian business hours. Plans depend on hours, desks and coverage. Get a staffing plan within 1 business day.