What a Virtual Assistant Costs Your Insurance Agency

The hour you're not selling

Nine o'clock on a Tuesday, and you're still at the desk. Not calling the two leads that came in that afternoon, not following up on the quote you sent last week. You're updating five renewal dates in the CRM, because nobody else is going to do it and the carrier portal only lets you export one client file at a time. That's real work. It also doesn't require a license, and it isn't the hour that pays you like a producer.

This is the hour an insurance agency virtual assistant exists to take off your desk, and most agency owners never actually price it out. They know, in a general way, that hiring help costs money. What they don't do is put a number next to the hour they're already spending, the one that disappears into the CRM instead of the phone.

That's the gap this guide closes. Not "you should get help," which every agent has heard a hundred times, but the actual arithmetic: what your own hour is worth by federal wage data, what a week of admin work costs at that rate, which tasks legally have to stay with a licensed agent, and which ones don't. Then what a virtual assistant paired with automation actually changes, and where the honest limits are.

Why agents keep doing this themselves

Ask an agency owner why they still do their own data entry, and you'll hear some version of three answers. First, "I'll hire when I'm bigger," which treats the cost of doing it yourself as free in the meantime, when it isn't. Second, "nobody else knows my book well enough," which is true on day one and stops being true the moment a process exists to hand off. Third, and the one agents say less often out loud: the CRM is messy enough that handing it to anyone feels riskier than just doing it yourself one more week.

All three are understandable. None of them make the hour free. The math in this guide doesn't ask you to believe hiring help is a good idea in the abstract. It asks you to look at what your own hour is already priced at, by the same government wage data that prices every other occupation in the country, and decide from there.

There's also a real, separate reason agents hesitate: client data. An insurance CRM holds health information, income details, and beneficiary data. Handing that to an outside person without a real access and compliance setup is a legitimate worry, not an excuse. We'll come back to that directly, because it's the part most VA services skip.

A messy CRM makes all three excuses worse. If your pipeline stages don't mean anything, if half your contacts have no notes attached, and if the same lead shows up three times under three spellings, handing that off to anyone feels like handing off a mess rather than a process. That's a real problem, but it's a different problem than the one this guide solves, and it's usually the reason a VA engagement stalls in the first month. A CRM that actually holds context on every contact, the way we cover in our guide to AI context in the CRM, is what makes delegation work instead of just relocating the mess.

What your own hour is actually worth

The Bureau of Labor Statistics tracks wages for insurance sales agents (SOC 41-3021) two ways: the Occupational Outlook Handbook, updated with May 2024 figures, and the more granular Occupational Employment and Wage Statistics survey, whose most recent published percentile breakdown is dated May 2023¹. Both agree on the order of magnitude, and both are worth knowing, because the spread between a new agent and an established one is wide.

The Occupational Outlook Handbook puts the current median at $60,370 a year². The more detailed OEWS survey, one plan year prior, reported a median of $59,080 a year, or $28.40 an hour, with a mean of $79,700 a year across the full range of agents, from brand new producers to top earners¹. The Department of Labor's O*NET database, which draws on the same underlying wage series, lists a 2025 figure of $62,280 a year³. Three separate government pages, three points in time, and the number moves in a narrow, consistent band. That's what a well sourced wage figure looks like.

The two BLS releases measure slightly different things, and it's worth knowing why they don't match exactly. The Occupational Outlook Handbook is the plain-language summary BLS updates most often, meant for someone deciding whether to enter an occupation. The Occupational Employment and Wage Statistics survey behind it is the underlying data collection, built from actual employer payroll reports across hundreds of thousands of establishments, and it's the source with the percentile breakdown. When a number matters to a business decision, checking both is the right habit, not a shortcut around either one.

Read that table by finding yourself in it, not by using the median as a universal stand in. A newer agent nearer the 25th percentile is still pricing their own hour at close to $21. An established agent nearer the 75th or 90th is pricing it at $40 to $65. The gap between what an hour of your time is worth and what a VA or automation costs only gets wider as you move up that table, which is the opposite of the intuition most owners have.

The math on a week of admin work

Here's the part most owners never sit down and do. Take the OEWS median, $28.40 an hour, and multiply it by whatever block of non-selling admin time you actually spend in a normal week. Eight hours is a reasonable middle estimate for an agent handling their own renewals, data entry, and recordkeeping, though your own number could be higher or lower. At eight hours, that's $227.20 a week spent on work priced at a producer's rate. Over a 50-week year, that's $11,360, not in cash you write a check for, but in selling time you never get back.

Now compare that hour to what it should cost. The Bureau of Labor Statistics separately tracks bookkeeping, accounting, and auditing clerks, the occupation whose entire job is "compute, classify, and record data to help organizations keep complete and accurate financial records"⁴, which is a near exact description of CRM entry and renewal processing. The median wage for that occupation was $49,210 in May 2024⁴, against $60,370 for an insurance sales agent in the same release². On a standard 2,080 hour work year, that's roughly $23.66 an hour for clerk-level work, against $29.02 an hour for a producer's time, both figures derived from the same BLS pay period.

That last row is worth sitting with. The BLS projects clerk-level recordkeeping roles to shrink over the next decade, not because the work disappears, but because software and automation absorb more of it every year⁴. Insurance sales agent roles, meanwhile, are still projected to grow². The work isn't going away. The question is only whether it stays priced at your rate, or moves to the rate it actually costs.

Run the same math at a few different weekly loads, using the OEWS median of $28.40 an hour, and the number moves fast. Five hours a week of admin work costs $142 a week, or $7,100 across a 50-week year. Eight hours costs $227.20 a week, or $11,360 a year, the scenario above. Twelve hours, which is not unusual during a renewal push or an enrollment period, costs $340.80 a week, or $17,040 a year. None of those totals include the leads that went uncalled while the hour was spent, which is the part that doesn't show up in any wage table at all.

What has to stay with you, and what doesn't

The Department of Labor's O*NET database lists the actual task set for insurance sales agents, and it splits cleanly into two categories³. The first needs a license and your judgment. The second is administrative, and it's exactly the category a trained VA or an automation can take over without touching anything that requires you personally.

Notice what isn't on the delegable list: talking a client into a decision, or making one for them. That line doesn't move. What moves is everything downstream of the decision, which is most of the paperwork that actually eats an evening.

Finding your own split doesn't require guessing. For two weeks, keep a running note, on paper or in a phone, of every task you touch and how long it takes. At the end of each day, mark each entry with an L for licensed judgment or a D for delegable admin. Most agents who actually do this, rather than estimate from memory, are surprised by how much of the D column there is, and by how much of it repeats in nearly the same form week to week. That repetition is exactly what a trained VA, backed by automation, is built to absorb.

Virtual assistant, AI automation, or a hire

Three different fixes get pitched for the same problem, and they aren't interchangeable. A human virtual assistant is good at judgment calls inside a defined process, like reading a renewal file and flagging what actually changed before it goes into the CRM. Automation is good at repeatable, rules-based work: sending a reminder on a fixed schedule, syncing a field between two systems, generating a first-draft follow-up. A full-time in-house hire makes sense once the volume is high and steady enough to fill a real job, which for most independent agencies is further down the road than owners assume.

Most agencies that get real hours back use a VA and automation together, not one instead of the other. Automation alone can't read a messy renewal file. A VA alone, without automation handling the repetitive parts, ends up doing manual work that software should have absorbed years ago. The combination is the point.

Say the honest part too: if your book is small and your admin load really is under an hour a week, none of this pencils out yet. The math in this guide is for the agent who already knows the answer is more than an hour, and just hasn't priced it.

The math changes again once you're running more than one book, which is the normal case for an FMO or IMO managing agents across several offices rather than the exception. A single agent weighing eight admin hours a week is one calculation. An organization with twenty producers each carrying a similar load is the same calculation twenty times over, and it's usually where the case for a shared VA and automation layer, instead of each agent solving this alone, becomes obvious. The same reactivation and follow-up automation that keeps a single agent's dead leads working, covered in our guide to database reactivation, scales the same way across a multi-office book.

The risk of handing this off carelessly

The reason agents hesitate on delegation isn't laziness, and it deserves a straight answer. An insurance CRM holds health information, income data, and beneficiary details. Handing a login to an outside contractor with no access controls and no signed Business Associate Agreement is a real exposure, not a paperwork formality, on anything touching Medicare, ACA, or general health coverage.

The fix isn't avoiding delegation. It's setting it up so the VA and any connected automation never touch raw protected health information in the first place. Ambrose, the platform Strategic AI Architects builds on, does this through a PHI gating function that passes coded, safe data between connected systems, so a CRM, a VA's task queue, or an automation gets what it needs to function without ever holding the raw record. Say this plainly: HIPAA compliant, never HIPAA certified, because no vendor can certify a practice as a whole.

If you're evaluating any VA service, insurance-specific or general, ask directly whether they'll sign a BAA and what data actually crosses to a human versus staying masked. A service that can't answer that clearly isn't ready for an insurance CRM, whatever else it offers. The same scrutiny applies to any tracking or analytics tool already running on your site; for enterprise level clients we will do a Business Associate Agreement covering exactly that, detailed in our BAA and NDA.

How we build the handoff

Strategic AI Architects builds VA + AI as a paired system, not a headcount you hire and hope integrates. A trained virtual assistant handles the judgment-level admin work, CRM hygiene, pipeline and task management, and routine tech support. Automation runs underneath it for the repeatable parts: reminders, data syncing, and appointment workflows, so the VA isn't doing by hand what software should already be doing⁵.

Pricing is scoped and priced on a call to the actual work required, not a bloated retainer for capacity you don't use, which is the same model the site states for custom builds generally: freelancer or agency alternatives for comparable custom work often run $5,000 to $25,000 or more upfront, delivered over weeks, against a scoped build delivered in days⁵. Everything runs on your own accounts, your own domain, and your own CRM. You own it, and it keeps working whether or not the relationship continues⁵.

None of this replaces you making the calls that require your license. It replaces the hours around those calls that don't.

The build itself runs through Ambrose, the AI platform behind the automation layer, and Claude Code on the engineering side. The strategy call is where the actual scope gets set: which tasks the VA takes, which run through automation, and which stay untouched because they belong to your license. Nothing ships until that scope is agreed, which is also why the price is a number you hear before work starts, not an estimate that grows once you're committed.

What the first month actually looks like

Handing off admin work fails when it starts with "here's my login, figure it out." It works when it starts with the two-week task log described above, turned into an actual plan before anyone touches your CRM. The order matters more than the speed.

Week one is the audit. You (or a VA working alongside you) run the two-week log, or a compressed version of it, to get a real list of what repeats: renewal reminders, data entry after a call, quote follow-ups, policy document filing. This is also when access gets scoped, meaning a VA or an automation tool gets exactly the permissions a task needs and nothing more, with PHI masked at the connection point rather than handed over whole.

Weeks two and three are a shadow period. The VA or the automation runs the delegable tasks with you reviewing the output before anything goes final, the same way you'd check a new hire's work before trusting it unsupervised. This is the part agencies skip when they're in a hurry, and it's the part that determines whether month two feels like relief or like cleanup.

By week four, the review step should be shrinking to spot checks rather than a full review of everything, and the tasks that were eating your evenings should be landing in the CRM before you ever see them, correct and on time. If they're not, that's a sign the task list from week one needs revisiting, not that delegation itself failed. Most of the friction in a bad handoff traces back to skipping the audit, not to the VA or the automation doing the work.

What you get back

The honest outcome isn't a promised income number, a lead count, or a ranking, because none of those are something any vendor can guarantee. What changes is concrete and measurable in your own week.

A clean CRM that's actually ready for automation to work on top of it. Selling hours back on the calendar, spent talking to the leads that already came in instead of catching up on the ones from last week. And a straight answer on what any of this costs, priced to your agency instead of guessed at.

Picture the same Tuesday from the top of this guide, six months later. The renewal dates synced into the CRM automatically when the carrier file came in. A VA flagged the two files that actually needed your eyes, with notes attached, before you sat down. The two leads from that afternoon got called at four o'clock instead of getting an apology text at nine. Nothing about your license changed. What changed is which hour got which task.

Questions agents ask