The salary line is not the comparison
Most agencies run the build-versus-buy decision on one number: the salary they would pay a local CSR against the monthly rate on an outsourcing quote. Those two numbers are not comparable, and the gap between them is not a rounding error.
A salary is one line on a payroll register. A managed remote seat is priced as a delivered service, so it already contains recruiting, payroll administration, benefits, workspace, supervision and coverage for absences. Setting one against the other compares a component to a bundle.
The fix is straightforward. Build the fully loaded annual cost of the in-house seat, convert both options to a cost per unit of serviced work, and be honest about the cases where outsourcing changes nothing. Below is the worksheet, the capacity math, and that list of cases.
Build the fully loaded cost of an in-house CSR
Every range below is illustrative and moves with state, metro, benefit design and carrier mix. Treat the numbers as placeholders and replace each one from a source you already have: your payroll register, your broker's latest benefits renewal, your AMS invoice, your lease.
The worksheet
| Cost line | Where to source your real number | Illustrative annual range |
|---|---|---|
| Base salary, experienced service CSR | Your last two offer letters | $45,000 to $65,000 |
| Payroll taxes (FICA, FUTA, SUTA) | Payroll provider tax summary | 8 to 11 percent of wages |
| Health, dental, vision, employer share | Benefits renewal | $5,500 to $15,000 |
| Retirement match at actual take-up | Plan document plus participation report | 2 to 4 percent of wages |
| Workers comp and other coverages | Policy schedule | $300 to $1,200 |
| Workspace: rent, utilities, furniture | Annual occupancy cost divided by seats | $2,000 to $7,000 |
| Hardware, amortized over three years | IT purchase records | $400 to $900 |
| Software: AMS named user, phone or CCaaS, email, e-signature, rater | Vendor invoices, per-user lines | $1,500 to $4,000 |
| Recruiting: agency fee or internal sourcing time | Last placement invoice | $3,000 to $10,000 |
| Ramp: first-year productivity loss | See the note below | 8 to 20 percent of first-year cost |
| Supervisor time, 2 to 4 hours per week | Manager's loaded hourly rate | $3,000 to $8,000 |
Add the lines up before you compare anything. For many agencies the year-one total lands between 1.3 and 1.6 times base salary, but calculate the multiplier rather than assuming it: benefits design alone can move it several points.
The lines people forget
- Ramp is a cost, not a warm-up. A CSR who is new to your AMS, your carrier appetites and your documentation standard usually runs at partial output for 6 to 12 weeks, even with prior agency experience. Price it as a share of first-year salary instead of treating it as free.
- PTO is a capacity cost, not a cash cost. Fifteen PTO days plus holidays removes roughly 10 percent of the seat's productive year. The cash still goes out the door. The endorsement queue still builds.
- Turnover is probabilistic, so budget it that way. If service roles in your market turn over every two to three years, then a third to a half of the recruiting plus ramp cost belongs in every annual budget, not only in the years someone resigns.
- Software seats usually do not disappear. Most agency management systems license per named user, and you want a named user for everyone touching a file so the activity log stays clean. That line follows the work, not the office.
What a managed remote seat covers, and what it does not
A managed seat is a bundle, and the comparison is only fair once you know what is inside it.
Usually inside the rate: recruiting and screening, employment and payroll administration, workstation and connectivity, day-to-day supervision, recorded quality reviews, SOP upkeep, and a cross-trained backup when the person is out.
Usually still yours:
- Named user licenses in the AMS, the phone system, carrier portals and any rater that bills per user.
- Process definition. Someone inside the agency has to decide how a certificate of insurance request or an endorsement is handled, and to what standard.
- Licensed activity. Quoting, binding and coverage advice sit with appropriately licensed staff regardless of where those staff work.
- Escalation ownership. A remote team member routes an unhappy insured to a named person, and that person is on your org chart.
The capacity math: cost per 1,000 policies
Cost per seat is the wrong denominator. Cost per unit of serviced work is the one that survives a leadership meeting.
Worked example for a personal lines book, with the assumptions stated so you can swap in your own:
- Book size: 1,000 active personal lines policies.
- Service touches per policy per year: 4. A renewal review, an endorsement or ID card request, a billing question, a document request.
- Average handle time including AMS documentation: 14 minutes.
- Productive hours per full-time seat per year: 1,500. That is 2,080 paid hours minus PTO, holidays, training, meetings and breaks.
Those assumptions give 4,000 transactions at 14 minutes, or about 933 hours of service work. That is roughly 0.62 of a full-time seat per 1,000 personal lines policies.
You cannot hire 0.62 of a CSR. Rounding up to one is where in-house cost quietly hides: you fund a full seat, absorb the idle third as "we have coverage," and never see it as a line item on any report.
Commercial lines runs heavier. Certificates of insurance, endorsements, premium audits, loss run requests and renewal data collection push both touch counts and handle times up, so plan on 1.5 to 2.5 times the personal lines hours per policy until you measure your own book.
With a denominator in hand, both options go in the same table:
| Metric | In-house seat | Managed remote seat |
|---|---|---|
| Annual cost | Fully loaded total from the worksheet | Quoted rate times 12, plus your per-user software |
| Productive hours per year | About 1,500, single point of coverage | Contracted hours, with a cross-trained backup |
| Smallest unit you can buy | One full seat | Often a partial or shared allocation |
| Cost per 1,000 personal lines policies | Annual cost times 0.62 | Annual cost times 0.62 |
| Time to first productive output | Search plus 6 to 12 week ramp | Ramp only, on a defined SOP |
The fractional-seat row is where most of the real difference lives. If your service workload is 0.6 or 1.4 seats, the in-house option forces you to buy 1 or 2, and you pay for the difference either in idle time or in a backlog you keep apologizing for.
Where outsourcing does not save money
Five situations where the math does not work. Better to find them now than four months into a contract.
- Under about half a seat of genuinely defined work. Scoping, documenting and reviewing output is a mostly fixed cost. Below roughly 15 to 20 hours a week of well-defined tasks, the management attention required usually exceeds the savings.
- Undocumented processes. If nobody can describe a workflow well enough to write it down, outsourcing will not fix it. You will pay someone to execute an undefined process, get inconsistent output, and conclude that outsourcing does not work. Document first, then decide.
- A licensed-capacity bottleneck. If the real constraint is that you have too few licensed people to quote and bind, a service seat relieves the wrong constraint. The legitimate version of this play is different: move the non-licensed servicing and administrative work off the licensed person's plate so their licensed hours go further. Be precise about which work you are moving.
- Roles that are physically local. Walk-in traffic, wet signatures, courier runs, community events and in-person carrier relationships do not transfer, and pretending otherwise produces a bad first 90 days.
- A horizon under six months. Ramp cost has to amortize over something. A short, hard-dated project spike is usually better handled with overtime or a temp than with a new operating model.
Make the comparison defensible
Three rules keep the analysis from being argued away in the meeting where you present it.
- Model year one and year three separately. Year one carries recruiting and ramp on the in-house side. Year three carries turnover probability. A single-year view flatters whichever option you already prefer.
- Hold scope constant. Write out the exact task list. If the local hire would also cover the front desk and the mail, either keep that in both columns or remove it from both.
- Count the revenue side once, or not at all. Hours returned to a producer or an account executive have real value, but apply that value to both options or to neither. Applying it to one is how build-versus-buy models get quietly rigged.
Where to start
- Fill the worksheet with real numbers this week. One hour with your bookkeeper and your payroll register replaces every range in this article with your actual cost.
- Time two workflows end to end. Twenty certificate of insurance requests and twenty endorsements, including the AMS documentation. You now have measured handle times instead of estimates.
- Count service touches on a random 100 accounts. Pull twelve months of activity from your management system, divide by 100, and you have a defensible touch rate for your own book.
- Write the SOP for the single workflow you would move first. If you cannot get it onto two pages, that tells you the sequence: document, then delegate.
- Run the break-even at equal scope. Fully loaded annual in-house cost divided by 12, against the quoted monthly rate plus your per-user software. If the two are close, decide on capacity flexibility and coverage, not on price.