In-house vs outsourced customer support: a real cost breakdown
A US support rep costs 1.25-1.4x base salary once benefits, tools, hiring, and churn are counted. A two-agent in-house desk runs roughly $115,000-$160,000 a year; the same pair outsourced at published rates of $12-$22/hour runs about $50,000-$92,000. In-house still wins for early-stage teams and complex products.
Start with the number that never appears in the job posting. A US support rep on a typical $39,000-$46,000 base salary actually costs $50,000-$64,000 a year once the widely cited 1.25-1.4x fully loaded multiplier (benefits, payroll taxes, tools, overhead) is applied. Run that out for a two-agent desk and the in-house vs outsourced customer support question stops being abstract: roughly $115,000-$160,000 a year in-house with hiring and churn included, against roughly $50,000-$92,000 for two dedicated full-time outsourced agents at the $12-$22 per hour most mid-market engagements are quoted at. That gap is real. It is also not the whole story, and pretending otherwise is how bad outsourcing decisions get made.
The salary is the smallest surprise
Anyone who has run a support budget knows the job posting number is a floor, not a total. Here is what stacks on top of a $42,000 base salary for one US-based agent:
- Payroll taxes and benefits. Employer FICA, health insurance, retirement match, and paid time off typically add 25-40% on top of base pay. This is where the 1.25-1.4x multiplier, popularized by MIT’s Joseph Hadzima and repeated in nearly every employer cost guide, comes from.
- Helpdesk and tooling. Zendesk’s published per-agent pricing runs from $19 to $115+ per month depending on tier, before add-ons like QA and workforce management. Budget $700-$1,500 per agent per year for a realistic mid-tier setup, plus phone, QA, and knowledge base tools if you buy them separately.
- Hiring. SHRM puts the average cost per hire at roughly $4,700 once job ads, screening time, and background checks are counted. Agency recruiters charge more.
- Ramp time. A new agent handles a fraction of normal volume in month one and usually does not hit full productivity until month three or four. You pay full salary the whole time. (Nobody budgets for month two. Everybody lives through it.)
- Churn. Contact center turnover runs 30-45% annually according to QATC benchmarks, and published estimates put the cost of replacing one trained agent at $10,000-$20,000 in rehiring, retraining, and lost productivity. On a two-person team, that means budgeting for roughly one replacement cycle most years.
- Management. Someone has to schedule, coach, review tickets, and cover when an agent is sick. On a small team that someone is usually a founder or an ops lead whose time is worth more than the line item suggests.
None of this is exotic.
It just rarely gets added up in one place before someone posts the job.
What outsourcing includes, and what it does not
A dedicated-agent outsourcing rate is closer to a fully loaded cost than a salary is. The hourly or monthly fee generally covers recruiting, payroll, benefits in the agent’s country, supervision, scheduling, basic QA, and replacement when an agent leaves. Churn still happens on outsourced teams, but the provider absorbs the rehiring and most of the retraining instead of you.
What the rate does not cover matters just as much:
- Product training. Nobody knows your products, policies, and edge cases out of the box. You still invest real hours building the knowledge base and training the team, especially in the first month.
- Escalations. Refund exceptions, legal threats, and VIP customers still need someone on your side with authority to decide.
- Oversight. A provider runs QA, but you should still read a sample of tickets yourself; a handful a week is enough, and you are checking for tone drift, not typos. Outsourcing the work is not outsourcing the accountability.
- Setup and minimums. Many providers charge onboarding fees and require minimum monthly commitments, so a very low hourly rate on paper can cost more than it looks for tiny volumes.
In-house vs outsourced customer support: a two-agent example
Here is the comparison for a two-agent operation handling email and chat, using the published ranges above. Your numbers will differ; the structure will not.
| Annual cost line | In-house (2 US agents) | Outsourced (2 dedicated agents) |
|---|---|---|
| Base pay | $80,000-$92,000 | Included in rate |
| Payroll taxes + benefits (25-40%) | $20,000-$37,000 | Included in rate |
| Helpdesk seats + tooling | $1,500-$3,500 | Often included or shared |
| Hiring (~$4,700 per hire, amortized) | $3,000-$9,000 | Provider absorbs |
| Turnover replacement (30-45% churn) | $10,000-$20,000 in a typical year | Provider absorbs |
| Management and QA time | 10-20% of a manager’s year | Reduced, not zero |
| Approximate total | $115,000-$160,000 | $50,000-$92,000 at $12-$22/hr |
At the low end of published offshore rates ($8-$15 per hour), the outsourced number drops further; at onshore US outsourcing rates ($28-$45+ per hour), it can exceed the in-house figure. The savings come from labor arbitrage and shared infrastructure, not magic.
Four things the spreadsheet can’t price
Cost is only half the decision. Four things deserve equal weight:
Control. In-house agents sit in your standup, hear the roadmap, and absorb context by osmosis. An outsourced team gets what you document. If your documentation is thin, an outsourced team will expose that within two weeks.
Brand voice. A good provider can match your tone with enough examples and feedback cycles. But “good enough” arrives faster in-house, because the feedback loop is shorter.
Scaling speed. Adding an in-house agent takes one to three months of hiring plus ramp. A provider with a bench can add trained coverage in weeks. This cuts the other way too: scaling down in-house means layoffs; scaling down a contract means a conversation.
Coverage hours. Two in-house agents in one time zone give you roughly 8x5 coverage. Evenings, weekends, and holidays either go unanswered or get covered at overtime cost. Distributed outsourced teams make 16x7 or 24x7 coverage a scheduling exercise instead of a hiring project.
When in-house wins
Outsourcing is the wrong answer often enough that the cases deserve their own section.
Early stage, low volume. Under roughly 15-20 tickets a day, founders and early employees should be answering support themselves. Those tickets are the cheapest product research you will ever get, and handing them off too early means losing the signal along with the workload.
Genuinely complex products. If resolving a typical ticket requires engineering context, regulatory judgment, or deep technical troubleshooting, the training investment for an external team may never pay back. Keep tier-2 in-house permanently and only consider outsourcing tier-1 triage.
Brand voice as the moat. Some brands, particularly in luxury and community-driven niches, win precisely because support feels like talking to the founder. If that is your differentiation, protect it and pay the premium.
Heavy integration with internal systems. When agents need access to admin panels, order databases, and internal Slack to resolve anything, the security and access overhead of an external team can eat the savings.
How to actually run the decision
Pull your last 90 days of ticket volume. Under 20 tickets a day: stay in-house, keep answering them yourself if you can, and revisit in six months. Above that and growing: price both columns of the table with your own numbers (real local salaries times 1.3, real tool costs, an honest estimate of whose time currently manages the queue), then let coverage hours and scaling direction break the tie. The spreadsheet frames the decision. It should not make it for you.