Service file · Research VA · Updated November 2026
Outsourcing desk research: hiring a research virtual assistant
A buyer's file for moving desk research (market and competitor scans, lead and prospect lists, web research and data collection, fact-checking, source summaries and survey admin) to a dedicated remote researcher: what goes in scope, the evidence standard to require, which data rules to set, who keeps the conclusions, what it costs against a local hire and which numbers to put in writing.
- Rate
- $6.50–$10/hr
- Full-time / month
- $1,040–$1,600
- First candidates
- About 2 business days
- Setup / recruitment fee
- None
File summary
per hour for a dedicated research assistant through our recruiting partner
a month for a part-time seat of 80 hours; $1,040–$1,600 for a full-time seat of 160 hours
recruitment or setup fee; nothing is paid before you choose a candidate
business days to the first candidates, usually (never a binding promise)
at least, linked and dated, next to every finding the assistant hands you
paid databases, list tools and survey software stay on your own bill and your own licence
In this file
- Define the seat
- Work packages: what moves and what stays
- Decision rights
- The evidence standard
- Data-collection rules
- A statement of work you can copy
- The work order for each task
- Engagement models compared
- Total cost of ownership
- Sizing the seat
- Deliverables and acceptance criteria
- Scope by buyer type
- Software and data sources
- Controls
- Service levels and KPIs
- Vendor questionnaire
- Acceptance test
- Onboarding plan: 30, 60, 90 days
- Risk register and exit
- Requesting candidates
- FAQ
Define the seat
A research virtual assistant is a dedicated remote assistant who answers business questions with sourced facts, inside your own tools. You buy a block of hours, not a report. In those hours the assistant turns your questions into search plans, collects information from the sources you allow, checks it against the original page, filing or report, and hands it back in a format you can act on: a one-page brief, a spreadsheet, a competitor matrix or clean CRM records. You keep the questions, the conclusions and the decisions.
Four neighbouring seats sit close to this one. Pick the one you are buying before you write the request, because "research experience" on a CV covers all of them.
| Seat | What it produces | This file? |
|---|---|---|
| Research assistant (desk research) | Sourced briefs, competitor matrices, prospect lists, data sheets, fact-checks, source summaries, survey tallies | Yes |
| Data analyst | Cleaned datasets you already own, dashboards, statistics, SQL queries | No: a separate, higher-priced seat; say so if the role mixes both |
| Lead generation or outreach rep | Calls, emails and booked meetings with the prospects on a list | No: see our cold calling file; the research seat builds the list |
| Market research firm | Panel surveys, focus groups, interview programmes, study design and interpretation | No: the assistant can run small surveys in your tool and tabulate answers, not design studies |
Write a one-line seat definition into the request, for example: "Research assistant for a B2B software company; weekly competitor tracking on eight named companies, monthly prospect lists of about 150 accounts from our customer profile, fact-checks for the blog; no outreach, no analysis of our own sales data."
Work packages: what moves to the assistant and what stays with you
Buy the seat by package. Each package names the work, the output and who signs off. Start with one recurring package that has a clear format, such as a weekly competitor check or a monthly prospect list, then add one-off questions once the standard holds.
| Work package | Typical tasks | In scope? |
|---|---|---|
| Market scans | Published reports, government data and trade press collected on a market; customer segments listed; price points in a category gathered; the inputs behind your market estimate laid out in a table | Yes: every figure with its source and date; the estimate itself is yours |
| Competitor tracking | Named competitors' websites, pricing pages, product changes, job posts, reviews and press checked on a schedule; a change log kept | Yes: public information only |
| Lead and prospect lists | Companies matching your written profile; decision-makers and titles checked; duplicates removed; recent news noted; records loaded into your CRM | Yes: under your personal-data rules (below) |
| Web research and data collection | Public information from websites, directories and official registers gathered into one consistent sheet, with a source column on every row | Yes: by hand, by licensed export or by official API, within each site's terms |
| Fact-checking | Statistics, quotes and claims in your drafts, decks and reports traced to the original source; each one marked confirmed, corrected or unconfirmed | Yes |
| Literature and source summaries | Searches in Google Scholar, PubMed or similar for studies on a business topic; plain-language summaries of what each paper says; an annotated reading list | Yes: summaries of what sources say, no conclusions of the assistant's own |
| Supplier and vendor comparisons | Suppliers, software or service providers found and compared against your criteria; published terms and quotes collected; open questions listed | Yes: the choice is yours |
| Survey admin | Surveys built in your tool from questions you approved, sent to your opted-in contacts, responses tracked, results cleaned and tabulated | Yes: question design and interpretation stay with you |
| Reporting | An end-of-day note and a weekly summary (see service levels below) | Yes |
| Strategic conclusions, investment, legal, financial or medical advice, contacting prospects, buying data or tools, primary research design | Anything that commits your money, your position or your name | No |
Decision rights: who gathers, who decides
The assistant gathers and checks; you decide. Put this table in the onboarding pack so the line is written down before the first task.
| Decision | Assistant | You |
|---|---|---|
| The question and what it is for | Restates it as a search plan; asks what is unclear before starting | Set the question, the decision it feeds and the time box |
| Which sources count | Proposes sources; flags weak or paywalled ones | Approve the source list and the banned list |
| What the findings mean | Summarizes what the sources say; shows gaps and conflicts | Draw the conclusion |
| Market size or forecast | Collects the published inputs, each with its source | Build or approve the estimate and its assumptions |
| Which personal data may go on a list | Follows your written rules; flags anything outside them | Decide, with counsel where needed |
| Contacting prospects or respondents | Prepares the list and notes; sends surveys only to contacts you approved | Decide who is contacted and how |
| Buying a report, database seat or list | Finds the options and their published terms | Approve and pay |
| Investment, legal, financial or medical questions | Collects and cites what sources say, labelled as source summaries | Decide with your qualified advisers |
The evidence standard
Research is only as good as the sources under it, and a neat summary can hide a weak one. Write this standard into the agreement and check against it every week in the first month.
- A source on every finding: each figure, quote and fact carries a link (or a document and page reference) and the date it was read. No source, no finding.
- Original over secondary: a statistic is traced to the body that published it (the agency, the company filing, the study), not to the blog post that repeated it.
- Two sources for key figures: the numbers a decision rests on are confirmed in two independent sources, or labelled "single source".
- Labels, not hedges: every item is marked as confirmed, estimate, single source or conflicting; conflicting sources are shown side by side, not averaged away.
- Answer first: a brief opens with the answer to your question, then the evidence, then the open points.
- AI is a tool, not a source: the assistant may use AI tools you approve to draft search terms or summaries, then checks every claim against the original and removes what it can't confirm. An AI answer never appears as a citation.
- Stop and ask: when sources conflict on something that matters, when the data sits behind a paywall you don't have, or when the task will run past the time box.
Data-collection rules
How information is collected matters as much as what is found. The rules below are general information for writing your scope, not legal advice; your counsel decides how the law applies to your lists and your markets.
- Website terms: the assistant follows each site's terms of use. Automated collection (scraping) is used only where the site's terms allow it and you approve the method; otherwise information is collected by hand, by export from a tool you license or by the site's official API. Logins, paywalls and blocks are never worked around.
- Paid databases: the assistant works under its own seat where your plan allows another user, and inside the vendor's licence terms on use and sharing. One paid seat is not shared between people against those terms, and downloaded data is used only as the licence allows.
- Personal data: prospect lists hold names, titles and business contact details, which are personal data under laws such as the GDPR in Europe and state privacy laws in the US, and commercial email to those contacts falls under CAN-SPAM in the US. You decide what may be collected, where it is stored, how long it is kept and how it is used; check the programme with counsel. The assistant keeps a source column so you can say where each record came from.
- Confidential material: your internal documents stay in your drive, and nothing confidential is pasted into AI tools or outside services you haven't approved.
A statement of work you can copy
Fill in the brackets and attach the result to your request and to the agreement.
Role: Research assistant (remote) for [company], a [industry] business selling to [customers] in [markets].
Hours: [80 / 160] hours a month, [days and hours], [time zone]; deliveries due by [daily / weekly cut-off].
Access: own users in [shared drive, CRM, research databases, survey tool, project tracker], with two-factor sign-in and only the rights the packages need.
Work packages in scope: [market scans, competitor tracking, lead and prospect lists, web research and data collection, fact-checking, source summaries, supplier comparisons, survey admin, reporting].
Out of scope: strategic conclusions, investment, legal, financial or medical advice, contacting prospects, buying data or tools, study design, scraping against a site's terms, any shared login.
Standards attached: evidence standard, approved and banned source list, personal-data rules for lists, output templates, decision-rights table.
Reporting: end-of-day note (tasks done, tasks open, questions waiting on you) and a weekly summary.
Commercials: hourly rate, monthly billing in arrears for hours worked, 30-day trial, notice period, replacement terms.
The work order for each task
The statement of work covers the seat. Each research task then gets a short work order, so an open request like "look into our competitors" can't turn into a month of hours. Keep it as a template in your tracker.
Question: one sentence, for example "Which companies sell [product] to [segment] in [market], and what do they publish about price?"
Decision it feeds: what you will do with the answer, so the assistant knows what matters and what doesn't.
Sources: which count (official sites, filings, registers, your paid databases) and which don't (forums, unsourced articles, AI answers).
Output: the format and columns, with an example of a good result.
Time box and due date: the maximum hours before checking in, and when you need it.
Engagement models compared
Businesses buy desk research in five ways. They combine: a research firm for one large study, with a dedicated assistant keeping the tracking current afterwards, is one way to combine them.
| Criterion | You or your team | Freelancer per project | Market research firm | Bought data list | Dedicated research VA |
|---|---|---|---|---|---|
| Cost model | No invoice; paid in senior hours | Per project or per hour | Per study, quoted | Per record or subscription | Flat $6.50–$10/hr |
| Recurring tracking | When there is time | Rarely | Usually not | Refresh depends on the vendor | Yes, on your schedule |
| Primary research | Rarely | Sometimes | Yes | No | Small surveys in your tool only |
| Fit to your question | High | Depends on the brief | High, at study scale | Generic fields | High, and checked against your rules |
| Who holds the knowledge | You | The freelancer | The firm's report | The vendor | Your seat, in your own files |
For the general trade-off between one dedicated person and a managed team, see dedicated team vs staff augmentation.
Total cost of ownership: local hire vs research VA
The closest US job title is market research analyst, a more senior role that also designs studies and interprets results. For list building and data collection the closer comparison is an administrative assistant. The table compares a year of each at the national median with a full-time remote seat.
| Line item (per year, full-time) | US market research analyst | US administrative assistant | Dedicated research VA |
|---|---|---|---|
| Pay or hourly fees | $78,760 median ($37.87 an hour) | $48,310 median ($23.23 an hour) | $12,480–$19,200 (160 hours × 12 months at $6.50–$10) |
| Employer payroll taxes | About $6,025 (7.65% FICA) | About $3,696 | None for you |
| Pay plus FICA | About $84,785 | About $52,006 | $12,480–$19,200 |
| Unemployment taxes and benefits | State and federal unemployment on top; benefits commonly about 30% of total compensation in US private industry | None for you | |
| Desk, computer, phone | You provide | The VA's own computer; budget any tool seats | |
| Recruiting | Job ads and your interview hours | No recruitment fee | |
| Databases, list tools, survey software | The same whichever model you choose; an extra seat may be needed for the assistant. | ||
If your research doesn't fill a full-time seat, the realistic comparison is the part-time seat: 80 hours a month costs $520–$800, or $6,240–$9,600 a year. Weigh it against the cost the table leaves out: the senior hours your founders, sales reps and marketers spend building lists and checking numbers today.
Sizing the seat
Size the seat from your own month. List every recurring research task, how often it runs and roughly how long it takes. The example is a B2B services company; the inputs are illustrations, not benchmarks.
| Item (example inputs) | Per month | Hours each | Hours a month |
|---|---|---|---|
| Prospect list of 150 accounts, checked and loaded to the CRM | 1 | 30 | 30 |
| Competitor check on eight companies, change log | 4 | 3 | 12 |
| Fact-check of a blog post or deck | 6 | 1.5 | 9 |
| One-page brief on an ad hoc question | 2 | 6 | 12 |
| Supplier or software comparison | 1 | 8 | 8 |
| Weekly summary and end-of-day notes | 4 | 1.5 | 6 |
| Seat to buy | About 77 hours a month: a part-time seat of 80 hours, with little room for a large one-off project | ||
Deliverables and acceptance criteria
Agree the format of each deliverable and what makes it acceptable. That turns "is the research good?" into a check you can run in minutes.
| Deliverable | Accepted when |
|---|---|
| One-page brief | Answer in the first paragraph; every figure linked and dated; open points listed; within the time box |
| Competitor matrix | One row per company, the columns you set, the date each cell was checked; changes since last time highlighted |
| Prospect list | Every record matches the written profile; no duplicates against your CRM; title and company checked; source column filled |
| Data sheet | One format throughout; required fields filled or marked "not found"; a source on every row |
| Fact-check log | Each claim marked confirmed, corrected or unconfirmed, with the original source and the exact figure or wording found there |
| Annotated reading list | Full citation, a plain summary of what each source says, its date and type (study, review, report); no conclusions beyond the sources |
| Survey tally | Response counts, cleaned answers, tables by the questions you set; raw export kept in your drive |
Scope by buyer type
Briefs before decisions
A new market, a supplier, a competitor's move: one-page briefs, answer first, with the evidence and the open points behind it.
Lists and account notes
Prospect lists from your profile, decision-makers checked, recent news noted, records loaded to the CRM so reps sell instead of searching.
Sources and fact-checks
Statistics and examples traced to the original before writing starts, and every published figure checked again before it goes out.
Products and suppliers
Product, keyword and review research and supplier comparisons; marketplace operations sit with an Amazon or Etsy seat.
Public records and data
Property data, permit histories and public records pulled into one sheet; the value and investment calls stay with you and your advisers.
Software and data sources
There is no standard research toolkit. Name your tools in the request so candidates can be matched to them, and test them in the interview. The names below are examples of tools candidates can be matched to, not endorsements.
| Area | Examples |
|---|---|
| Market and industry | Google Trends, government statistics portals, public company filings, Crunchbase, report libraries you subscribe to |
| Competitors and web | Semrush, Ahrefs, Similarweb, LinkedIn, review sites, web archives |
| Prospects | LinkedIn Sales Navigator, Apollo.io, ZoomInfo, Hunter; your CRM (HubSpot, Salesforce, Pipedrive) |
| Literature | Google Scholar, PubMed, Semantic Scholar |
| Surveys | Google Forms, Microsoft Forms, SurveyMonkey, Typeform |
| Organizing and reporting | Google Sheets, Excel, Airtable, Notion, Google Docs, Slides, PowerPoint |
| AI assistants | Only the tools you approve, for drafts the assistant then checks against original sources |
Controls
Require
- Own users in every tool, with two-factor sign-in and only the rights the packages need
- Database seats under your account and the vendor's licence terms
- Written evidence standard and approved source list
- Written personal-data rules for every list
- Work stored in your own drive and CRM
- A signed confidentiality agreement covering your data and findings
- A time box on every task
- Access removed the same day the engagement ends
Refuse
- Shared logins to paid databases
- Scraping against a site's terms, or getting around logins and paywalls
- Findings without a source, or AI answers cited as sources
- Prospect data copied into personal sheets or chat apps
- Summaries presented as investment, legal or medical advice
Service levels and KPIs to put in writing
Measure for 30 days, then set targets from your own numbers. Each measure comes from a place you can already check.
| Area | Measure | Data source | Review |
|---|---|---|---|
| Delivery | Tasks delivered by the due date and within the time box | Your tracker | Weekly |
| Source quality | Share of sampled findings whose source exists, says what the brief says and is correctly dated | Your spot-check log | Weekly, then monthly |
| List quality | Records outside the profile, duplicates and bounced or wrong contacts reported by your team | CRM, sales feedback | Monthly |
| Tracking | Scheduled competitor checks run on time; change log updated | Change log | Weekly |
| Rework | Tasks sent back for missing sources or wrong scope | Your tracker | Monthly |
| Reporting | End-of-day note every working day; weekly summary on the agreed day | Your inbox | Weekly |
Keep business outcomes out of the agreement. Deals closed, market share or a launch's success depend on your decisions, not on the person who gathered the facts.
Vendor questionnaire: ten questions
- Is the assistant dedicated to my business, and what research experience will candidates have?
- Are candidates tested on a sourced research task from my own work before I meet them?
- Do I interview and choose?
- What is the hourly rate, and what moves it within the range?
- How do I move between 80 and 160 hours, or add hours for a large project?
- Are there recruitment or setup fees, or an advance payment?
- What are the trial period, notice period and replacement policy?
- Will the assistant work only under named users in my tools and within each database's licence?
- How are my data and prospect records protected on the assistant's side?
- What are the backup for internet and power outages and the cover during leave?
Our outsourcing contract checklist covers the clauses behind these answers.
Acceptance test before you choose
Ask each shortlisted candidate for a short work sample built on your own business:
- Competitor table: the published plans and prices of five competitors you name, in a table with a source link and date for each cell. Where a price isn't published, the right answer says so.
- Fact-check: three statistics from a draft of yours. Trace each one to its original source and mark it confirmed, corrected or unconfirmed.
- List build: twenty companies matching a written profile, with one decision-maker each and a source column.
- Summary: a long industry report cut to five points for a busy founder, answer first.
- Judgment trap: include a request to "find the personal mobile numbers" of the list, or to pull data from behind a login. The right answer stops and asks.
Score accuracy, sourcing and judgment. Ask each candidate to walk you through the sources they trusted and the ones they rejected.
Onboarding plan: 30, 60, 90 days
- Week 1, access and standards: users and seats set up before day one; the evidence standard, source list, data rules, templates and two or three past research examples shared; a first small task due within one working day.
- Weeks 2–4, one package under full review: one complete brief and one recurring list or check; you review the sources, not only the summary, and note what to change.
- Day 30 review: baseline numbers for the service levels; seat size checked against the tally; the standing rules written into an SOP.
- Days 31–60, recurring work: competitor tracking, list building and fact-checking on a schedule; spot checks weekly.
- Days 61–90, full scope: ad hoc briefs, supplier comparisons and survey admin added; spot checks monthly where accuracy has held; targets written into the agreement.
Risk register and exit
| Risk | Control | Owner |
|---|---|---|
| A decision made on a wrong or unsourced figure | Evidence standard; two sources for key figures; weekly spot checks | You and the assistant |
| Data collected against a site's terms or a database licence | Written collection rules; own seats; methods approved by you | You |
| Personal data used outside your rules | Written list rules checked with counsel; source column; storage only in your CRM | You |
| Open-ended tasks eating the month | Work order with a time box on every task | The assistant |
| Access misused or lost | Own users with two-factor sign-in; no payment access; access removed on the last day | You and the provider |
All work lives in your drive, CRM and tools, so nothing has to be handed back. At the end of an engagement, ask for a handover note (tasks open, recurring checks and their schedules, sources in use), remove access the same day and close any database seats you no longer need.
Requesting candidates
Requests sent through this site go to our recruiting partner, a managed VA agency:
- Send the scope: the research you do today, how often, the tools and the hours, using the statement of work above if it helps.
- Review the shortlist: candidates with research experience, screened and tested, usually in about 2 business days.
- Interview and choose: nothing is paid before you choose; billing is monthly in arrears for hours worked; an assistant who isn't a fit is replaced at no extra cost.
The trade-offs: an agency costs more per hour than a freelancer you find yourself, in exchange for the search, screening and replacement; the first month needs your time to write the standards and review sources; after the 30-day trial a notice period applies.
FAQ
What does a research virtual assistant do?
Desk research for your business, done remotely in your own tools: market and competitor scans, lead and prospect lists built to your customer profile, web research and data collection into spreadsheets, fact-checking, summaries of reports and published studies, supplier comparisons and survey admin. Every finding comes back with its source and date. The questions, the conclusions and the decisions stay with you.
How much does a research virtual assistant cost?
Through our recruiting partner, $6.50–$10 an hour: about $520–$800 a month for 80 hours or $1,040–$1,600 a month full-time (160 hours), billed monthly in arrears for hours worked, with no recruitment or setup fee. Paid databases, data tools and survey software stay on your own bill.
Is a research VA cheaper than an in-house market research analyst?
Per year, yes, though the seats differ. The US Bureau of Labor Statistics put the median pay of market research analysts at $78,760 a year ($37.87 an hour) in May 2025, before payroll taxes and benefits. A full-time remote research seat costs $12,480–$19,200 a year at $6.50–$10 an hour. An analyst who designs studies and interprets results for strategy is a more senior hire; the research VA gathers, checks and organizes.
Can a virtual assistant do market research?
Desk market research, yes: collecting published reports and public data, mapping competitors and prices, summarizing industry news and building the tables behind a market estimate, with each figure sourced and dated. Primary research such as panel surveys, focus groups or interview programmes is a job for a research firm; the assistant can still draft questions for your approval, run surveys in your own tool and tabulate the answers.
Can a research VA build lead lists for my sales team?
Yes: companies and contacts that match your written customer profile, with titles checked, duplicates removed and notes added, loaded into your CRM. Collecting and using personal contact data is covered by laws such as CAN-SPAM in the US and the GDPR in Europe, so you decide what may be collected and how it is used, and check the programme with counsel. Contacting the leads is a separate outreach seat.
Will a research VA scrape websites?
Only where the website's terms allow it and you approve the method. By default the assistant collects public information by hand, exports from tools you license, or uses a site's official API. It never gets around logins, paywalls or blocks, never shares one paid seat between people against the vendor's terms, and stops and asks when a site's terms are unclear.
How do I check the quality of the research?
Write a source standard into the agreement: a link and access date next to every finding, two independent sources for the figures a decision rests on (or a single-source label), estimates marked as estimates, and conflicts shown rather than hidden. Then spot-check a sample of sources every week in the first month and every month after that.
Can AI tools replace a research VA?
AI tools are quick at first drafts, search terms and summaries, and they can be wrong or out of date. In this file an AI answer never counts as a source: the assistant may use AI tools you approve to speed up the draft, then checks every claim against the original page, report or filing and removes what can't be confirmed. Don't paste confidential material into AI tools you haven't approved.
Will the research VA give investment, legal or medical conclusions?
No. The assistant collects and summarizes what sources say, with citations. It doesn't recommend investments, interpret laws or contracts, or draw medical conclusions from studies. Those calls belong to you and your qualified advisers: your financial adviser, your attorney or a qualified clinician.
How long does it take to get a research VA working?
First candidates usually arrive in about 2 business days. Plan about 30 days to small sourced briefs and one recurring list under full review, 60 days to competitor tracking and fact-checking on a schedule, and 90 days to full scope with lighter spot checks. With our recruiting partner you can end the engagement immediately during the first 30 days; a notice period applies after that, and an assistant who isn't a fit is replaced at no extra cost.