In-house vs. outsourced breakeven calculator

The wage is the job, not a round number

An outbound appointment-setting seat and an inbound service desk are different published occupations. The US Bureau of Labor Statistics puts the median telemarketer (SOC 41-9041) at $17.04 an hour and the median customer service representative (SOC 43-4051) at $21.53 -- nine thousand dollars a year apart. Pricing an outbound floor at the customer-service wage, or an inbound desk at the telemarketer wage, prices the wrong job. The direction selector on the calculator above swaps between the two published medians rather than asking you to type one.

Above the agents sits a supervisor, and here is a boundary worth stating plainly: BLS publishes no "call center manager" occupation. The closest published series is first-line supervisors of office and administrative support workers (SOC 43-1011), median $33.41 an hour. That is a general office-and-administrative supervisor figure, not a contact-centre one, and relabelling it "call center manager" would attribute a number to an occupation BLS did not measure -- which is the quiet way a sourced figure becomes a fabricated one.

The loaded-cost multiplier on top of both wages is a slider on the calculator for a reason: it is your assumption, not a BLS figure, and neither this page nor any fact it cites publishes one. Move it and watch the fixed cost move with it -- that transparency is the entire difference between this tool and a single invented total.

The corrected worked example

Two seats at 40 hours a week, outbound at $17.04 base with a 1.30 load multiplier, a 0.25 supervisor FTE at $33.41, $150 per seat per month in tools, 60% annual turnover at $1,500 a hire, 10 area codes dialed, 30% shrinkage, 20 dials per productive hour and 2.0% booked per dial, against a $200 quoted price: agents $92,152, supervisor $22,585, tools $3,600, hiring $1,800, DNC fee $410 -- a fixed annual cost of $120,547. That funds 1,165 appointments a year at $103.49 each against $200 bought.

The band: below 603 appointments a year, buying at $200 each is cheaper than carrying that fixed cost. Above 1,165 appointments a year, these two seats are already maxed out and a third seat -- with its own fixed cost -- is what buys more capacity, not a faster dialer. Between the two is the only volume range where this floor beats buying, and it is a range, not a point.

Same wages, same rent, same dialer -- the hire decides it

The published spread across ccdocs' 11 agent scorecards is measured PER CONTACTED HOMEOWNER: 1.2% to 5.3%, a 4.5x spread. That ratio does not transfer to a per-dial axis, because contact rate partly offsets booking rate -- scaling it would overstate the per-dial spread. Computed directly from the same scorecards' total-call and appointment-booked counts instead, never scaled: the lowest published setter booked 2 of 291 dials (0.687% per dial) and the highest booked 43 of 1544 dials (2.785% per dial) -- a 4.05x spread. No agent and no campaign is named; these are the extreme values among the 11 published scorecards, not an identified person.

Run through the same annual dial pace as the worked example above, the worst published setter's capacity sits below any breakeven this calculator can produce at a reasonable seat count -- that hire can never win, no matter how much volume is thrown at them. The best published setter clears it with room to spare. Same wages, same rent, same dialer: the decision is decided by who you can hire, not by the wage line.

Both denominators, because either one alone misleads

The calculator's default booking rate is sourced to one fact, and that fact's own methodology requires both readings to travel together: 3.6% of contacted homeowners booked an inspection (210 of 5,772 contact calls); 2.0% of all dials did (212 of 10,794 calls). Source: apps/airoofing/files/10k_agent_scorecards.md (per-agent contact and appointment counts, summed over the 11 published scorecards) plus apps/airoofing/files/10k_winning_vs_losing.md and apps/airoofing/roadmap.md (corpus totals: 10,794 roofing calls, 212 booked).

The per-contact figure alone flatters, because it hides that most dials never reach a person. The per-dial figure alone damns, because it charges the agent for disconnected numbers and no-answers neither side controls. The calculator defaults to the per-dial reading because "dials per productive hour" is the input it asks for -- replace both the activity and the rate with your own numbers rather than trusting either denominator to describe your operation.

Where this model breaks down

• The load multiplier is your assumption. BLS publishes wages, not a loaded-cost figure for payroll taxes, benefits and overhead -- nothing on this page or cited by it sources one.
• The booking rates are outbound roofing, one estate, Feb-Mar 2026. They do not transfer to inbound, B2B or another vertical. Replace the dial pace and booking rate before reading anything into your own result.
• This is annual arithmetic and hides seasonality. A fixed cost is paid every month, including a storm business's quiet quarter -- this model does not size that.
• Ramp is not priced. The days from a signed agreement to a first booked appointment are not measured yet, so this calculator prices day one at full productivity rather than guessing at a ramp.
• Headcount is an integer. Capacity is a step function -- the ladder panel on the calculator shows the step rather than smoothing it into a curve.
• The $200 default is one roofing offer, carrying its own 20-appointment minimum and $4,000 floor -- not a generic rate. The field is editable so you can enter a rival quote instead.
• Deliberately excluded and not modelled: state telemarketer registration, surety bonds, errors-and-omissions coverage, facilities, and the cost of an idle seat in a slow week. None of those has a published figure this page can cite.
• The supervisor occupation is named honestly. BLS publishes no "call center manager" series; first-line supervisors of office and administrative support workers (SOC 43-1011) is the closest published code, and calling it anything more specific would be a fabricated attribution.

The Do Not Call registry fee

An outbound operation dialing its own list pays the National Do Not Call Registry access fee: the first 5 area codes are free, then $82 each, capped at $22,626 a year (16 CFR 310.8(c)). The fee resets every fiscal year, so treat the figure above as current rather than permanent, and this is the registry access cost only -- it is not the cost of scrubbing software, a dialer or a compliance review, none of which this calculator prices. It is zero for an inbound seat, which places no outbound calls against the registry.

Where this fits

• Build a call center -- the page whose first FAQ asks the question this calculator answers with a number instead of a range.
• Pricing -- the quoted per-appointment side of this comparison, and the source of the calculator's editable default price.
• Agent headcount calculator -- sizes the floor this tool prices; use it first if you do not yet know how many seats to model.
• Call center cost estimator -- prices the outsourced side alone, across the three common vendor pricing models.
• Outbound call center -- the buying-side alternative this arithmetic sits beside.