The real cost of an in-house seat

Almost every in-house-versus-outsourced comparison we are shown starts from the wrong number. Someone takes the salary of an agent, multiplies by headcount, and compares it to a vendor's seat rate. The vendor rate looks expensive, the conversation ends, and the real comparison never happens.

Salary is the smallest line in a seat. Here is what the rest of it is made of.

The costs that sit on top of salary

• Employer contributions and statutory benefits. Fixed, predictable, and usually 10–20% on top depending on jurisdiction — the easy part.
• Recruitment. Sourcing, screening, interviewing and onboarding one agent costs real money and real management hours. Multiply by your attrition rate, because you pay it again every time someone leaves.
• Attrition itself. A seat that turns over twice a year is paid for three times: the leaver's notice period, the vacancy, and the replacement's ramp.
• Training and ramp. Four weeks of paid training is four weeks of salary producing nothing, plus a trainer's time. If you ramp ten people you have also lost a supervisor for a month.
• Floor, power, internet, and telephony. Per seat, per month, whether the seat is filled or not.
• Licences. CRM, helpdesk, dialler, QA tooling, workforce management. Often the second-largest line after salary and almost always left out of the in-house estimate.
• Supervision. A team lead per ten to fifteen agents, a QA analyst per twenty to thirty, a trainer, a scheduler. These roles are invisible in a headcount plan that only counts agents.
• Idle capacity. This is the one that decides the answer.

Idle capacity is the whole argument

An in-house team is sized for peak, or it misses peak. If your busiest hour needs twenty agents and your average hour needs twelve, you employ twenty and pay eight of them to wait for eight hours a day. That is not bad management — it is what happens when your only lever is hiring.

Expressed differently: your effective cost per handled contact is not salary divided by hours worked. It is total cost divided by contacts actually handled. Teams routinely discover their real utilisation is 55–65% once shrinkage, breaks, training, meetings and absence are counted honestly.

An outsourced pod moves that risk. Agents can be shared across campaigns, shifts can be built around your volume curve rather than around a standard working day, and you buy handled volume rather than attendance.

How to run the comparison properly

Build both sides as an annual cost per handled contact:

• Take twelve months of total in-house spend on the function — every line above, not just payroll.
• Divide by contacts actually handled in the same period.
• Ask the vendor for the same figure at your real volume, with your real SLA, at your real peak-to-average ratio.

Then check three things before believing either number. Does the in-house figure include supervision and licences? Does the vendor figure include QA, reporting and the ramp period, or are those quoted separately? And is the vendor pricing per seat or per handled contact — because per-seat pricing hands the idle-capacity risk straight back to you.

When in-house still wins

Outsourcing is not always cheaper, and a partner worth having will tell you so. Keep it in-house when volume is genuinely low and stable, when the work needs deep proprietary judgement that takes a year to build, when regulatory constraints make third-party access disproportionately expensive, or when the function is your product rather than support for it.

The honest test is simple. If your cost per handled contact is already competitive at your real utilisation, you do not have an outsourcing problem. If you cannot calculate that number at all, that is the finding — and it is worth more than any quote.