Virtual Assistants for Insurance Leads: Cost and Risk

Yes, but only for the parts of the call that don’t require a license, and that line is easier to cross than most agencies think. A virtual assistant can dial your list, confirm basic eligibility, and book a callback window without touching insurance licensing law at all. The moment that same VA starts describing what a plan covers, comparing one carrier to another, or asking someone to apply for a specific policy, they’re doing something the NAIC’s own model act calls “soliciting,” and that requires a producer license in the state the lead lives in, regardless of what country the VA is calling from.

That distinction rarely comes up in the online conversation about hiring a VA to call insurance leads, which agencies usually start looking into for the same reason a local hire is hard to find in the first place. That conversation mostly focuses on cost: a Filipino telemarketing VA runs $8.50 to $14.50 an hour according to VA MASTERS’ own published rates, a fully managed VA through MyOutDesk starts at $1,988 a month, and either one is dramatically cheaper than a $44,770-a-year US-based hire, per the Bureau of Labor Statistics’ median wage data for customer service representatives. All of that is true and worth knowing. None of it answers the question that actually determines whether the arrangement is safe to run: what is the VA allowed to say once someone picks up the phone.

What “Hiring a Virtual Assistant” Actually Means

“VA” covers three different arrangements that get lumped together in agent Facebook groups, and they carry different cost structures and different amounts of oversight you have to build yourself.

The first is a freelance marketplace hire: you post a job on Upwork, OnlineJobs.ph, or a similar platform, interview candidates directly, and manage payroll, training, and performance yourself. This is the cheapest option on an hourly basis and the one with the least built-in structure. You are the HR department.

The second is a specialized VA staffing service, like VA MASTERS, which recruits and places a dedicated telemarketing VA with you for an hourly or monthly rate, handling sourcing and often offering a replacement if the placement doesn’t work out. You still manage the VA’s day-to-day work and script, but someone else handled the hiring funnel.

The third is a fully managed VA agency, like MyOutDesk, which bundles recruiting, vetting, payroll, benefits, HR administration, onboarding, and a dedicated account manager into one flat monthly rate. You get less choice over exactly who you’re assigned and pay more for it, but you’re not running a mini HR operation for a single remote hire.

None of these three is the same thing as outsourcing to a BPO or call center, where you’re leasing a team’s capacity rather than hiring a specific individual, and the vicarious liability questions that come with a third-party vendor apply differently than they do to a person working under your direct supervision. A VA, in any of the three forms above, is functioning much closer to an employee or a direct contractor of your agency, which is exactly why the licensing question below lands on you more squarely than it would with an arm’s-length vendor.

What a Calling VA Actually Costs in 2026

The rate you’ll see quoted depends heavily on which of the three arrangements above you’re looking at, and the spread is wide enough that “a VA costs $10 an hour” and “a VA costs $2,000 a month” can both be true statements about the same job.

Notice the pattern across every hired-human arrangement in that table: you pay for the hours or the seat, not for the outcome. A VA who dials for eight hours and connects with six people gets paid the same as one who connects with sixty. That’s not a criticism of the workers; it’s just how hourly and flat-rate staffing is structured, and it’s the single biggest mechanical difference from a per-minute-of-talk-time billing model.

The Math Nobody Runs: Cost Per Connected Minute

Here’s a worked example using only the vendor-published numbers above, not a performance benchmark, to show why “cost per hour” and “cost per outcome” tell you different things. The same per-minute logic drives the live-agent-versus-AI cost comparison elsewhere on this site; the math below applies it specifically to a hired VA instead of a live in-house agent.

Say your agency needs roughly 500 minutes of actual qualifying conversation with leads in a given month, a reasonable number for a few hundred calls that connect and run a few minutes each. A full-time VA through MyOutDesk costs $1,988 that month whether those 500 minutes happen or not, because you’re paying for the seat, not the connected time. VA MASTERS’ offshore telemarketing VA runs $1,360 to $2,320 a month full-time under the same logic.

A managed AI caller on TheAffordableAI’s Single Account bills only for connected talk time: 500 minutes at $0.20 a minute is $100, plus the $200 monthly platform fee, for $300 that month, or $275 at the $0.15 bulk rate. No charge accrues for voicemail connections or failed dial attempts, according to the plan’s own published billing terms.

The gap widens further at higher volume, because the VA’s flat rate doesn’t move whether you need 200 connected minutes or 2,000 in a given month, while the AI’s cost scales with actual usage in both directions. That also means the AI option can look worse than a VA in a genuinely slow month with very little call volume, since the VA at least represents fixed, predictable capacity. Neither of these facts, by itself, answers the licensing question, which is the part that actually determines whether either arrangement is safe to scale.

The crossover point is worth working out for your own agency rather than taking the 500-minute example on faith. At the $0.20-a-minute Single Account rate, connected talk time would have to reach roughly 9,940 minutes in a month, about 166 hours of pure connected conversation, before the AI’s usage-based cost caught up to MyOutDesk’s flat $1,988. That’s an enormous volume for qualification calls that typically run two to four minutes each; most agencies running a few hundred leads a month land nowhere near it. The comparison flips even further in the AI’s favor at the $0.15 bulk rate, which pushes the crossover past 13,000 minutes. This is simple algebra on published rate cards, not a claim about how much volume your agency actually generates, so run it with your own numbers before deciding anything.

You can build a version of this yourself with a dialer and a VA. Plenty of agencies do, and for some volumes it’s the right call. It’s worth pricing both against your own numbers rather than either one on faith.

The VA’s cost line stays flat across all three rows because that’s the nature of a seat-based rate. The AI’s line only overtakes a flat-rate VA at a volume most single-agent operations won’t hit; the calculation looks very different for an Agency-plan operation running thousands of connected minutes across 20+ agents, where the flat-rate comparison would need to be against several VA seats, not one.

Edge Cases Worth Planning For

A few situations complicate the clean comparison above, and it’s worth thinking through them before you commit to either arrangement.

A VA calling leads in multiple states. The NAIC model act’s definitions apply to insurance solicited in “this state,” meaning the licensing question tracks where the lead lives, not where the VA sits. A VA working a Texas list and a Georgia list in the same shift needs the same qualification-only discipline for both, since both states’ producer licensing statutes are built from close variations of the same model language, even though Georgia’s own telemarketing statute adds separate consent requirements on top of the licensing question.

Medicare leads specifically. CMS’s TPMO disclaimer requirement and current Scope of Appointment rules apply regardless of whether a VA, an AI caller, or a licensed agent makes the call, and a VA who starts walking a Medicare lead through plan differences is both soliciting without a license and skipping a required federal marketing disclosure in the same breath. Route Medicare qualification calls to a strictly eligibility-and-scheduling script, full stop.

Part-time VA arrangements. A VA working 20 hours a week instead of 40 cuts the monthly cost roughly in half using the same hourly rate, which narrows the gap with a per-minute AI caller at lower volumes but doesn’t touch the licensing question at all. A part-time VA who solicits is exactly as exposed as a full-time one.

Mixing a VA with an AI caller. Some agencies run both: an AI caller for the first-touch qualification wave and a VA for warm relationship management with an existing book of clients. That split can work well precisely because it keeps the highest-volume, most repetitive, most license-sensitive work on the system built to run the same script every time, and reserves the VA’s time for work that genuinely benefits from a person who knows the client.

The Licensing Problem Nobody Mentions

Every blog post ranking for “insurance agency virtual assistant” talks about cost, English proficiency, and time zones. Almost none of them mention that the content of the call itself can trigger a state producer licensing requirement, independent of who’s paying whom or what country the call originates from.

The NAIC’s Producer Licensing Model Act, the template most state insurance codes are built from, defines the relevant terms plainly. Section 3 states: “A person shall not sell, solicit or negotiate insurance in this state for any class or classes of insurance unless the person is licensed for that line of authority in accordance with this Act.” The definitions in Section 2 do the real work:

Read that “solicit” definition again: attempting to sell, or asking or urging a person to apply for a particular kind of insurance from a particular company. That’s a lower bar than most agencies assume. A script that has a VA say “based on what you told me, a lot of people your age go with a Medicare Advantage plan instead of Supplement, want me to walk you through why” is urging a person toward a particular kind of insurance. It doesn’t matter that the VA never quotes a premium or names a carrier by name if the pitch itself is doing the urging. A script that instead says “I just need to confirm you’re on Medicare Part A and B and find a 15-minute window for one of our licensed agents to call you back” is neither selling, soliciting, nor negotiating under any of the three definitions above.

The commission rule in Section 13 catches a second, quieter problem. Agencies commonly pay a VA a small bonus per booked appointment that converts to a sale, on top of the hourly rate, to keep the VA motivated. If that VA isn’t a licensed producer, that bonus is compensation “for selling, soliciting, or negotiating insurance” the moment it’s tied to whether the lead actually bought something, and Section 13 bars both sides of that transaction, the agency paying it and the VA accepting it. A flat hourly rate, or a bonus tied to a non-sale metric like calls completed or appointments booked regardless of outcome, sits outside that rule.

The mistakes that actually show up in practice

A handful of patterns account for most of the real exposure agencies walk into with a calling VA, and none of them require bad intent.

• Writing a script that sounds like qualification but functions like a pitch. “A lot of people in your situation end up going with X, want me to explain why” is urging, not qualifying, even if it never mentions a specific premium.
• Letting a motivated VA improvise past the script. A VA who genuinely wants to help a confused caller will sometimes explain a benefit unprompted. That instinct is the exact behavior the script has to train against, not reward.
• Structuring pay around outcomes without checking Section 13. A per-sale bonus feels like normal sales-team incentive design and is exactly what the commission rule was written to catch when the person earning it isn’t a licensed producer.
• Assuming a staffing agency’s compliance language covers this. A VA staffing company’s marketing page focuses on recruiting, payroll, and replacement guarantees, not on your state’s producer licensing statute. That’s your review to do, not theirs.
• Treating “the VA never quoted a price” as the whole test. Solicitation under the model act’s definition doesn’t require a price quote. Urging someone toward a particular kind of insurance from a particular company is enough on its own.

Turnover, Ramp Time, and the Quality You Actually Get

Cost per hour and licensing exposure aren’t the only variables. A VA also has to be trained, has to be managed, and can leave. The commonly cited “40 to 45 percent annual call center turnover” figures that circulate in marketing blog posts rarely name a source or a methodology, so this guide won’t repeat them. What’s actually measurable from a primary source: the Bureau of Labor Statistics’ JOLTS report put the quits rate for professional and business services, the broad government category VA staffing sits inside, at 1.8 percent for July 2026, a level of 401,000 quits that month. That’s one month’s rate across a sector far wider than phone-based lead qualification alone, so treat it as a general signal about labor market churn in adjacent job categories, not a VA-specific attrition statistic.

What that number doesn’t capture, and what agencies actually feel, is ramp time. A new VA needs to learn your product, your CRM, your disqualification criteria, and the exact line between “confirming eligibility” and “urging someone to apply,” and that training investment resets every time a placement doesn’t work out. A managed agency like MyOutDesk builds a replacement guarantee into its flat rate specifically because this is a known cost of the model; a freelance marketplace hire usually doesn’t come with that cushion built in.

If you want to hear what the qualification-only version of this actually sounds like before you decide anything, there’s a demo call on the homepage; putting your own number in takes less time than reading the next section.

None of this means a VA does worse work than a person on your own payroll would. It means the labor-market and management overhead that comes with any human calling role, sourcing, training, script drift, and eventual replacement, applies here the same as it would to an in-house hire, just distributed across a vendor relationship instead of your own HR function.

Virtual Assistant vs Managed AI Caller, Side by Side

The core difference isn’t intelligence or even cost, it’s consistency of scope. A well-trained VA can absolutely stay inside the qualification-only lane every single call. The risk is that “well-trained” has to be re-earned after every hire, every script update, and every long shift, and a single call where a VA drifts into describing plan benefits is the call that creates exposure. A managed AI caller’s script doesn’t have good days and bad days; it runs the version you configured, which makes the qualification-versus-solicitation line something you set once at the platform level instead of re-enforcing with every new person who picks up a headset.

When a Virtual Assistant Is Still the Right Call

None of this makes a virtual assistant the wrong tool everywhere. A VA makes sense when the work benefits from an ongoing human relationship at a volume that doesn’t justify full automation: a bilingual VA who knows a specific book of Spanish-speaking clients personally and can handle nuanced questions a script can’t anticipate, someone managing your calendar, inbox, and light client communication alongside occasional calls, or a VA doing entirely non-phone work like carrier paperwork, data entry, or CRM cleanup where the licensing question never arises because no solicitation is happening at all.

Where it gets risky is routing high-volume, first-touch, script-driven lead calling, exactly the job most agencies hire a calling VA to do, to a role that’s billed by the hour regardless of outcome and has no structural guardrail against a well-meaning VA trying to be more helpful on a call than the license allows. If you already have a VA doing this work today, the fix usually isn’t firing them. It’s auditing the actual script against the definitions above and moving anything that sounds like urging someone toward a specific plan to a licensed agent’s side of the handoff.

Where TheAffordableAI Fits

I want to be direct about what changes and what doesn’t when you swap a VA for a managed AI caller: the licensing rule is identical either way. What a well-built AI calling setup does is make the qualification-only scope the default, instead of something you have to keep re-training into a rotating cast of hires. The full feature list covers how each piece works together; the six that matter most for this comparison are below.

Where to Start This Week

You don’t need to fire anyone or rip out a working process to close the gap here. Start with these three things.

1. Pull the actual script your VA is using, not the one you think they’re using, and check every line against the model act’s definitions above. Anything that describes benefits, compares carriers, or urges a specific application gets moved to a licensed agent.
2. Check how any performance bonus is structured. If it’s tied to whether a lead converts to a sale and the VA isn’t a licensed producer, that arrangement runs directly into Section 13’s commission rule.
3. Run the cost-per-connected-minute math for your own volume, using your actual monthly minutes rather than the illustrative 500 used here, before deciding whether a flat-rate VA or a per-minute AI caller fits your call volume better.

A virtual assistant can be the right, affordable answer to “I can’t hire fast enough.” What it can’t do is turn a licensing requirement into an optional one just because the person on the call is a contractor instead of an employee, or because the call originates outside the state, or the country, the lead lives in. Write the script first. Decide who’s allowed to say what. Then decide whether a VA, an AI caller, or both is the right way to run it.