“What does outsourced customer support cost?” depends entirely on which of three pricing models you are buying — per-seat, per-ticket, or a dedicated agent — and vendors rarely make the differences plain, because the confusion is profitable. This guide explains all three models honestly, benchmarks the underlying labour markets in the US, India and the Philippines with sources and dates, and shows the math on a worked 5-seat example.
A dedicated remote support agent costs $6–$8 per agent hour through a managed provider — $1,056–$1,408/month full-time (176 hours). BPO alternatives are priced per seat, per hour or per ticket, and vary widely with contract size, channel mix and hours of coverage — ranges differ enough between vendors that any single number would mislead, so this guide explains the structures instead. The US baseline: a customer-service-representative median wage of $44,770/year (BLS OEWS May 2025), about $4,850/month loaded. Local markets: India CSR average ₹342,279/year (PayScale, n=560); Philippines CSR average ₳232,308/year (PayScale, n=826).
The Three Pricing Models, Explained
| Model | How you pay | Best for | The catch |
|---|---|---|---|
| Per seat (BPO) | Monthly fee per staffed seat, often with minimum seat counts and term commitments | Large, stable volume; 24/7 coverage across shared shifts | The agent in the seat rotates; your product knowledge walks out with every rotation |
| Per ticket / per transaction | A rate per resolved ticket, chat or call minute | Spiky or unpredictable volume | The incentive is closing tickets fast, not solving them well; complex tickets get deflected |
| Dedicated agent | One named agent, full- or part-time, at an hourly/monthly rate | Product-heavy support where context compounds | You manage the queue and quality — it is your agent, not a black-box service |
| Per interaction (AI-assisted) | A rate per AI-handled interaction, with human escalation billed separately | High-volume tier-0 intake in front of a human team | Containment-rate assumptions drive the quote — audit them against your actual ticket mix |
The structural difference matters more than the sticker price. Per-seat and per-ticket models sell you capacity; a dedicated agent is headcount — someone who learns your product, your tone and your edge cases, and gets better every month. For support where every ticket is nearly identical, capacity is fine. For support that touches a real product, context wins. The same three structures price call-center outsourcing in India and BPO contracts generally — the labels on the proposal change, the math does not.
The US Baseline, Fully Loaded
The BLS OEWS May 2025 median for Customer Service Representatives (SOC 43-4051) is $44,770/year. With employer benefits averaging about 30% of total compensation (BLS ECEC), a conservative loaded figure is 1.3× salary ≈ $58,200/year, or ~$4,850/month per agent — before recruiting, training, software seats or the turnover that support teams are notorious for.
India vs Philippines: the Labour Benchmarks
Both markets are deep, English-fluent and BPO-mature; the self-reported data (directional, not definitive) sits close together. PayScale India (read August 2026) reports a CSR average of ₹342,279/year across 560 profiles, 10th–90th percentile ₹194,000–₹604,000 — roughly $2,000–$6,400/year at ₹95/$. PayScale Philippines reports a CSR average of ₳232,308/year across 826 profiles — roughly $4,100/year at ₳56/$ — with a wide percentile spread that reflects the market’s range from entry voice work to team leads.
The practical read: cost alone does not pick the country. The Philippines has the deeper bench for US-consumer voice work; India has the deeper bench for technical, B2B and tools-heavy support. Zedtreeo recruits from India — stated plainly so you can weigh this guide accordingly.
One market dynamic affects both countries and deserves a line in any honest cost guide: attrition. High-volume support floors everywhere run elevated turnover, and every departure costs you re-training and a quality dip whether or not it shows on an invoice. The structural fix is the same one that fixes context loss: agents who own a named queue for a named client, with a visible skills path (tier-1 → tier-2 → QA or team lead), stay dramatically longer than agents rotating anonymously across accounts. When you evaluate any provider — us included — ask two questions: what was your agent retention on dedicated seats last year, and what does the growth path for my agent look like? The answers predict your year-two experience better than any rate card, because a $6/hour seat you re-train three times costs more than a $8/hour seat you keep.
Five Factors That Move the Rate
- Channel mix — chat and email price at the base of the band; voice, especially US-consumer voice, prices higher everywhere.
- Coverage hours — a single business-hours agent is the base case; true 24/7 needs a rota (see the 24/7 CS squad bundle for how that is actually staffed).
- Tier-2 depth — agents who can read logs, reproduce bugs and escalate precisely sit above pure tier-1 profiles.
- Tools — Zendesk, Intercom, Freshdesk and HubSpot experience is matched at shortlist; niche stacks narrow the pool.
- Languages — additional languages beyond English add a premium and shrink the candidate pool.
What 24/7 Coverage Actually Takes
The most common budgeting mistake in support outsourcing is pricing “24/7” as if it were a feature toggle. The arithmetic is unforgiving: a week has 168 hours and a full-time agent works about 40, so one continuously staffed seat needs a minimum of four agents on a rotating shift pattern — and in practice more, once weekly offs, leave and training are covered. That is why a vendor quoting a suspiciously cheap 24/7 line is usually selling you a shared bench: your tickets join a pool that other clients’ tickets also draw from, and the “dedicated” part quietly disappears outside business hours.
The honest ways to buy round-the-clock coverage, in ascending cost order: follow-the-sun with async channels (one or two dedicated agents on shifted hours, with chat/email SLAs that tolerate a gap); core team + overnight overflow (dedicated agents own the day, a shared or AI layer catches the night); and a true dedicated rota, which is a team purchase, not a seat purchase — the 24/7 CS squad bundle prices exactly that structure honestly. Match the model to what your night-time volume actually is: most SMB queues do not justify a full rota, and a well-set expectation (“replies within 4 hours overnight”) costs a fraction of pretending otherwise.
The Hidden Costs of Per-Ticket Contracts
Per-ticket pricing looks like the fairest model on paper — you pay for exactly what gets handled. The hidden costs live in the incentive structure. Speed over resolution: when revenue is per closed ticket, the rational vendor behaviour is to close fast, and reopened tickets often bill again — you can pay twice for one problem. Deflection creep: complex tickets that do not fit the rate card get bounced back to “out of scope” queues your own team ends up working. Definition games: what counts as one ticket (a thread? a contact? a resolution?) is where per-ticket contracts are won and lost — read that clause twice. Knowledge walkout: because no specific agent owns your queue, the product knowledge your tickets generate accrues to nobody; every quarter starts from the same baseline. None of this makes per-ticket wrong for high-volume, low-variance queues — it is genuinely the right tool there — but it explains why teams with product-shaped support migrate to dedicated agents after a year of watching the meter.
The Metrics That Keep Quality Honest
Whichever model you buy, instrument the same five numbers from day one, because they are where support quality actually lives. First response time (FRT) per channel — the number customers feel most. Resolution time split by tier, so a hard-ticket backlog cannot hide inside a fast-easy-ticket average. Reopen rate — the single best fast-close detector; anything drifting up while resolution time drifts down is a red flag. CSAT on resolved tickets, sampled consistently rather than cherry-picked. Escalation rate to your internal team — it should fall month over month as product knowledge compounds; if it does not, you have bought capacity, not capability. With a dedicated agent these metrics sit inside your own helpdesk analytics, which is precisely the point: quality assurance against your rubric and your ticket sample, not a vendor’s quarterly slide. Run the QA loop weekly for the first month (a five-ticket review takes twenty minutes), then relax to fortnightly once reopen rate and CSAT stabilise.
Where AI Fits (and Where It Does Not)
AI deflection has genuinely changed the denominator: well-implemented bots now absorb much of the repetitive tier-0 volume, which shrinks the ticket count but raises the average difficulty of what reaches a human. That shifts the economics toward fewer, better agents with product depth — exactly the dedicated model — and away from large rotating benches priced per ticket. We wrote the full analysis, with sourced deflection figures, in the AI customer support cost guide.
Channel Mix Economics: Voice, Chat and Email
The channel mix drives the rate more than any other single factor, because the three channels consume agent time differently. Voice is strictly serial — one agent, one caller — and demands the strongest spoken-accent fit, which is why US-consumer voice prices at the top of every market’s band. Chat allows limited concurrency (an experienced agent works two to three conversations at once), which is why chat-heavy queues get more effective coverage per seat than the hourly rate suggests. Email is fully asynchronous and batches well, making it the natural overnight channel in a follow-the-sun setup — and the most forgiving place for a new agent’s first two weeks. The practical implication for budgeting: state your actual ticket split by channel when collecting quotes. A “support agent” rate quoted without a channel mix is a number without a denominator, and the vendors who ask about your mix before quoting are the ones pricing honestly.
Onboarding: the First 30 Days on the Queue
The steady-state economics above arrive after an onboarding curve, and budgeting that curve honestly avoids the classic month-two disappointment. A realistic dedicated-agent ramp: week one, shadowing — the agent reads your macros, knowledge base and past tickets, and drafts replies your team approves before sending; week two, supervised live handling on the easiest ticket class with a same-day review loop; weeks three and four, expanding ticket classes with spot-check QA instead of full review. By day 30 a well-matched agent typically owns the tier-1 queue solo, with escalation paths doing the rest. Two things accelerate the curve more than anything else: a current knowledge base (even a rough one — the agent will improve it as they go, which becomes an asset you keep), and a named internal buddy for product questions during the first fortnight. The free 5-day trial maps onto the start of this curve deliberately — you see week one on your real queue before any money moves.
A Worked Example: the 5-Seat Team
A company running five US support agents at the loaded median (~$4,850 each, ~$24,250/month) moves the team to five dedicated remote agents at $6–$8/hour ($5,280–$7,040/month for all five), keeping the team lead and QA layer unchanged. The direct comparison saves $17,200–$18,970/month — roughly 71–78%, inside the 70–85% band we publish on the Rate Index. The honest caveat: month one carries training overhead, and voice-heavy queues may price above the band’s floor — budget the transition, not just the steady state.
Quality, QA and Data Handling
Support agents touch customer PII by definition, so the security posture is not optional. Zedtreeo engagements run under NDA with least-privilege access to your helpdesk and order systems, operated by LegelpTech Outsourcing Pvt Ltd, an ISO 27001:2022 certified company, with PCI- and HIPAA-aware workflows where the queue requires them. QA stays yours: your rubric, your ticket sampling, your CSAT targets — a dedicated agent is inside your quality system, not behind a vendor’s quarterly report.
How Zedtreeo Places Support Agents
Tell us the channel mix, the tools and the coverage window. You get a pre-vetted shortlist within 48 hours — every candidate has passed live written-English and scenario tests as part of the 6-stage standard — then a free 5-day trial on your real queue before any money moves. Start at hire remote customer support, see the customer support representative role page for rate detail, or add front-desk coverage with a virtual receptionist.
Frequently Asked Questions
How much does it cost to outsource customer support to India?
A dedicated full-time agent runs $1,056–$1,408/month ($6–$8/hour) through a managed provider. BPO per-seat and per-ticket contracts vary too widely with volume and terms for one honest number — get the structure clear before comparing quotes.
Per-seat, per-ticket or dedicated — which should I choose?
High-volume, low-variance queues suit per-seat or per-ticket capacity. Product-heavy support where context compounds suits a dedicated agent. Many teams run both: a bot plus dedicated humans for everything the bot cannot close.
India or the Philippines for support?
The Philippines has the deeper bench for US-consumer voice; India for technical, B2B and tools-heavy queues. The labour-cost difference is modest — pick on queue type, not price. Zedtreeo recruits from India.
Can agents cover US business hours?
Yes — Zedtreeo agents are shift-aligned to your business day. For true 24/7, the CS squad bundle staffs a proper rota rather than stretching one agent.
Will AI replace outsourced support agents?
AI absorbs repetitive tier-0 volume and makes the remaining human queue harder. The economics now favour fewer, better agents over large rotating benches — the full math is in our AI support cost guide.
How is customer data protected?
NDA on every engagement, least-privilege helpdesk access, and an ISO 27001:2022 certified operating company, with PCI- and HIPAA-aware workflows where required.
Can I start with one agent?
Yes — most teams start with one full-time or part-time agent (minimum 4 hours/day) and scale after the trial proves the model on their own queue.
How fast can an agent start?
Shortlist in 48 hours, then a free 5-day trial on your real tickets. Most placements are handling the queue solo inside two weeks.
Sources
- BLS OEWS May 2025 — Customer Service Representatives (43-4051), median $44,770
- PayScale India — CSR salary (n=560; read August 2026)
- PayScale Philippines — CSR salary (n=826; read August 2026)
- BLS Employer Costs for Employee Compensation — the ~30% benefits load
Methodology. US figures are the BLS OEWS May 2025 published medians; the loaded cost applies the BLS ECEC ~30% benefits share as a conservative 1.3× multiplier. India and Philippines figures are PayScale self-reported data taken without adjustment, with sample sizes disclosed. Currency conversions use single illustrative rates (₹95 = $1, ₳56 = $1, August 2026) and will drift with FX. BPO pricing structures are described qualitatively because contract-dependent ranges vary too widely to quote honestly. Zedtreeo’s $6–$8/hour band is the published starting range for support placements on the Rate Index. Reviewed August 2026.