How to Reduce Customer Service Costs in Ecommerce (Without Cutting Quality)

Most ecommerce owners don't have a customer service cost problem. They have a repeat-inquiry problem that shows up as a cost. The same questions arrive every day, "where's my order?", "can I return this?", "do you have it in blue?", and every one of them costs you an agent's time or your own evening.

This guide is about reducing that cost without making the service worse. We'll work through which inquiries to automate first, the levers that actually move your cost per contact, and the UK agent-cost context behind the number. The industry benchmarks are all here too, sourced and dated, but as evidence for the tactics, not as the point.

Quick answer: To cut ecommerce customer service costs, automate the repeat, low-judgement inquiries first ("where's my order?", returns, refund eligibility, stock questions), which are usually 40-60% of inbound. Move those to an AI agent that reads live order data and can act on it, keep your people for the cases that need judgement, and raise first contact resolution so issues don't come back. Labour is 60-70% of support cost, so the saving comes from resolving repeat work end to end, not from squeezing handle time.

This guide compiles 2024-2025 industry data on customer service costs. We'll cover cost per contact by channel, labour costs, technology investments, and the economics of automation. All data is sourced and put in context.

The Three Levers That Actually Move the Number

Before the benchmarks, the shape of the problem. Customer service cost is driven by three things, and they're worth fixing in this order:

1. Automate the repeat inquiries. The high-volume, low-judgement questions ("where's my order?", returns, refund eligibility, stock) are usually 40-60% of inbound. Resolved properly, they leave your team entirely.
2. Cut the labour you spend per resolution. Labour is 60-70% of total support cost, so anything that resolves an inquiry end to end, rather than handing it to a person, moves the number more than any other change.
3. Raise first contact resolution. Every issue that comes back a second time is paying twice. Lifting FCR cuts cost and lifts satisfaction at once.

Everything below is the evidence behind those three levers, and the figures you'd use to work out what they're worth on your store.

A quick caution on benchmarks, since this guide is full of them. Use them to find the gap, not to set the target. They tell you whether your costs are competitive, where to invest, and what efficiency gains are realistic. They don't tell you whether your service is good (cost is not quality), what your customers value, or whether you're solving the right problems. Competing on cost alone is a race to the bottom.

Cost Per Contact: The Core Metric

Cost per contact (CPC) measures the fully loaded cost of handling one customer interaction. It includes labor, technology, overhead, and management.

The formula:

CPC = (Total operational costs) / (Total contacts handled)

Total operational costs include:

• Agent salaries and benefits
• Supervisor and management salaries
• Technology costs (phone systems, CRM, chat platforms)
• Facilities (rent, utilities for contact center space)
• Training and quality assurance
• Overhead allocation

Overall Benchmarks

The industry average cost per contact across all channels is $2.70 to $5.60 according to recent call center benchmarking studies.

However, this range masks significant variation:

Gartner's 2024 research found that live channels (phone, live chat, email) cost an average of $8.01 per contact, while self-service channels (websites, mobile apps) cost approximately $0.10 per contact.

That's an 80x cost difference between live and self-service interactions.

The cost-per-call metric hit a five-year high in 2024, driven by:

• Increased agent compensation (tight labor market)
• Higher technology costs (AI platforms, security tools)
• More complex customer issues (simple queries moved to self-service)

Cost by Channel: The Real Breakdown

Different channels have dramatically different economics. Understanding these differences is critical for channel strategy.

Phone: The Premium Channel

Average cost per call: $4.27 to $8.01

Phone remains the most expensive channel for several reasons:

• Agents handle one call at a time (no concurrency)
• Calls require immediate response (no async efficiency)
• Complex issues escalate to phone (higher skill requirements)

Phone is expensive, but it's still necessary. Some issues require synchronous, empathetic communication. The key is ensuring only appropriate issues reach phone agents.

Live Chat: The Efficiency Middle Ground

Average cost per chat: $4.24 (similar to phone with low concurrency)

At first glance, chat appears no cheaper than phone. But the economics change with agent concurrency.

The rule of thumb: Chat costs approximately 33% less than phone when agents handle 3 simultaneous conversations.

The math:

• 1 chat = 2x the time of 1 phone call (per customer)
• 1 agent handling 3 chats simultaneously = 67% efficiency gain
• Result: $2.84 per chat vs. $4.27 per call

Chat's advantage is concurrency. Phone's advantage is resolution speed.

Email: Lower Cost, Higher Hidden Costs

Average cost per email: $3.81

Email appears cheaper than phone or chat. But there's a catch: resolution complexity.

Studies show:

• Fewer than 10% of emails can be answered fully without recourse to other channels
• Email chains often require 3-5 messages to resolve one issue
• Total resolution cost often exceeds phone when you count the full chain

Email works well for:

• Low-complexity questions with documented answers
• Asynchronous issues (order confirmations, status updates)
• Customers who prefer written communication

Email doesn't work well for:

• Issues requiring clarification (long back-and-forth)
• Emotionally charged situations (tone gets lost)
• Time-sensitive problems

Social Media: The Complexity Wild Card

Average cost per social contact: $3.64

Social media appears to be the cheapest live channel. But again, there's hidden complexity.

More than 50% of social media requests require completion via another channel (usually phone or email).

The true cost calculation:

Social media issue = $3.64 (social) + $4.27 (phone follow-up) = $7.91 total

Social media is valuable for brand presence and public resolution, but it's not a cost-saving channel when you account for full resolution.

Self-Service: The 80x Cost Advantage

Average cost per self-service interaction: $0.10

Self-service includes:

• Knowledge base articles
• FAQ pages
• Interactive troubleshooting tools
• Mobile app help sections

The economics are simple: build once, serve infinitely.

The problem: Only 9% of customers report solving their issues completely via self-service (Gartner, 2024).

The 91% who can't self-serve escalate to more expensive channels, often frustrated by failed self-service attempts.

The implication: Self-service is cost-effective only when it actually resolves issues. Poor self-service increases total cost. This is the gap an AI agent fills that a static FAQ page can't: it doesn't just point at the return policy, it reads the live order and answers "where's my order?" with the real tracking number.

Which Inquiries to Automate First

This is where the channel data becomes a plan. Not every inquiry is worth automating, and the ones to start with aren't a guess, they're the ones that repeat, have a documented right answer, and don't need a human's judgement.

For a typical ecommerce store, the inbound queue sorts into three tiers.

Tier 1, automate first (high volume, low judgement). These repeat daily and have one correct answer that lives in your order data or your policies:

• "Where's my order?" / tracking and delivery status
• Return and exchange requests
• Refund eligibility ("can I still send this back?")
• Shipping options, cut-off times, and delivery windows
• Stock and "do you have it in [size/colour]?" questions
• Order changes before dispatch (address, item swap)

On most stores these are 40-60% of everything that comes in, and "where's my order?" alone is often the single biggest line. They're also the ones that drag your evenings out, because they arrive at all hours and the answer is sitting right there in the system. An AI agent that reads live order data resolves these end to end. This tier is where the cost comes out.

Tier 2, assist rather than fully automate (medium judgement). Product fit advice, "which one should I buy?", multi-item troubleshooting. An AI agent can draft the answer and pull the relevant detail, but a person often wants to glance at it before it goes. The saving here is handle time, not full deflection.

Tier 3, keep human (high judgement, high emotion). Complaints, edge-case refunds outside policy, anything where someone's upset or money is in dispute. Automating these to save a few pounds costs you the relationship. Leave them with a person, and give that person more room by clearing Tier 1 off their plate.

The discipline is to automate Tier 1 ruthlessly and resist the temptation to push into Tier 3. Most of the saving is in Tier 1 anyway, so there's no need to reach for the cases that need a human.

Cost Per Contact by Industry

Different industries have different cost structures due to issue complexity, regulatory requirements, and labor market dynamics.

Industry benchmarks (all channels):

Industry Average CPC Range Retail $2.70 - $4.50 Lower complexity, high volume E-commerce $2.50 - $4.00 Digital-native, self-service focused Financial Services $5.00 - $8.00 Regulatory requirements, security needs Healthcare $6.00 - $9.00 HIPAA compliance, complex issues Telecommunications $8.00 - $12.00 Technical troubleshooting, high complexity Insurance $5.50 - $8.50 Claims processing, regulatory overhead SaaS/Technology $6.00 - $10.00 Technical support, product complexity

Why the variation?

Low-complexity industries (retail, e-commerce):

• Simple, repetitive questions
• Standardized processes
• Lower agent skill requirements
• Higher automation potential

High-complexity industries (telecom, healthcare):

• Unique, technical issues
• Regulatory compliance requirements
• Higher agent skill and training needs
• Lower automation applicability

Your industry matters. Compare your costs to your industry peers, not to the overall average.

Labour Costs: The Largest Line Item

Agent compensation typically represents 60-70% of total customer service costs. This is lever two, and it's the reason resolving an inquiry end to end beats every other efficiency tweak. Understanding labour market benchmarks is essential for budgeting.

UK agent cost in context

The detailed salary tables below are US figures (the cited research is US-based), but for a UK or Ireland store the shape is the same and the order of magnitude is what matters. A UK customer service agent typically sits around £24,000-£28,000 base salary. The fully loaded cost, once you add employer National Insurance, pension, holiday cover, training, tools, and a share of management time, runs roughly 1.3-1.5x that:

£26,000 base × 1.4 = ~£36,400 fully loaded
£36,400 ÷ 1,800 productive hours ≈ £20/hour effective cost

That per-hour number is what every "where's my order?" reply is quietly billed against. Clear a few hundred of those a month off an agent's plate and the saving is real money, not a rounding error. It's also why the cheapest support hire is the inquiry that never reaches a person at all.

Customer Service Representative Salaries (2024)

National averages (United States):

• Median hourly wage: $20.59 (U.S. Bureau of Labor Statistics, May 2024)
• Average hourly wage: $19.48 (Indeed)
• Annual salary range: $17,400 (lowest) to $55,500 (highest)
• Typical annual salary: $35,600

Entry-level compensation:

• Less than 1 year experience: $15.47/hour average
• Entry-level annual: $31,000 - $32,000

Geographic variation (highest-paying states):

• Alaska: $35,949/year average
• California: $48,804/year average
• Massachusetts: $48,154/year average
• San Jose, CA (metro): $48,254/year average

Geographic variation (lowest-paying states):

• Mississippi: $21,196/year average

That's a 70% difference between highest and lowest state averages.

Industry-specific pay premiums:

The highest-paying industries for customer service representatives (BLS, May 2023):

Industry Annual Mean Wage Premium vs. Median Support Activities for Water Transportation $94,790 166% Aerospace Product/Parts Manufacturing $70,620 95% Iron and Steel Mills $69,030 90% Motor Vehicle Manufacturing $67,320 86% Railroad Rolling Stock Manufacturing $67,120 86%

Industries with technical or safety-critical products pay 2-3x the median wage for customer service roles.

Manager Compensation

Customer service manager salaries:

• Entry-level (0-1 years): $97,259/year
• Intermediate (1-2 years): $97,800/year
• Senior (2-4 years): $99,064/year
• Specialist (5-8 years): $100,507/year
• Expert (8+ years): $103,258/year

Manager compensation is relatively flat across experience levels, suggesting the role requires domain expertise more than tenure.

Fully Loaded Labor Costs

Salary is only part of total labor cost. The fully loaded cost includes:

Benefits and taxes (typically 30-40% of base salary):

• Health insurance
• Retirement contributions
• Paid time off
• Payroll taxes
• Workers' compensation insurance

Training costs:

• Initial onboarding (2-6 weeks)
• Ongoing training (product updates, soft skills)
• Quality assurance coaching
• Attrition replacement costs

The calculation:

Fully loaded cost = Base salary × 1.35 to 1.50
Example: $35,600 salary → $48,060 to $53,400 fully loaded

Per-hour calculation for CPC:

$50,000 fully loaded annual cost ÷ 2,080 working hours/year = $24.04/hour
Add 15% for non-productive time (breaks, training) = $27.65/hour effective cost

This is the hourly cost number to use in cost per contact calculations.

Technology Costs: The Hidden Infrastructure

Technology costs have increased significantly in 2024 as organizations invest in AI, omnichannel platforms, and security infrastructure.

Core Platform Costs

Contact center platform (per agent/month):

• Basic cloud phone system: $40-$75
• Omnichannel platform (phone, chat, email): $100-$150
• Enterprise with AI features: $150-$250

CRM and ticketing:

• Basic help desk: $20-$50/agent/month
• Full CRM with service module: $75-$150/agent/month

Knowledge base and self-service:

• Basic FAQ platform: $500-$2,000/month (flat fee)
• AI-powered knowledge base: $5,000-$15,000/month

Workforce management:

• Scheduling and forecasting: $50-$100/agent/month

Quality assurance and analytics:

• Call recording and scoring: $30-$75/agent/month
• Speech analytics and sentiment: $100-$200/agent/month

Total technology cost per agent: $250-$800/month typical range

For a 100-agent contact center:

• Annual technology costs: $300,000 to $960,000
• Per-agent annual technology: $3,000 to $9,600

AI and Automation Costs

AI platforms are increasingly common but add complexity to cost modeling.

AI support platforms:

• Basic rule-based AI: $500-$2,000/month
• AI support (mid-market): $5,000-$15,000/month
• Enterprise AI platform: $25,000-$100,000/month

Conversation AI (real-time agent assist):

• Agent assist for 100 agents: $10,000-$30,000/month

Knowledge AI (answer suggestions):

• AI knowledge base: $5,000-$20,000/month

The ROI calculation is critical here. AI costs are justified only if they reduce labour costs or improve outcomes. The pricing model matters as much as the headline number: per-resolution and per-seat tools climb with the volume they handle, so the saving can leak straight back out. A flat monthly rate decouples the bill from volume, which is the model we use, and you can see ours on the pricing page.

Calculating Your Cost Per Contact

Here's how to calculate your own CPC and compare it to benchmarks.

Step 1: Identify Total Costs (Monthly)

Labor:

• Agent salaries (fully loaded): $______
• Supervisor salaries (fully loaded): $______
• Manager salaries (allocated): $______

Technology:

• Contact center platform: $______
• CRM/ticketing: $______
• Knowledge base: $______
• Quality assurance tools: $______
• AI/automation platforms: $______

Facilities:

• Rent (allocated to contact center): $______
• Utilities: $______

Other:

• Training and development: $______
• Telecom/internet: $______

Total monthly costs: $______

Step 2: Count Total Contacts (Monthly)

• Phone calls: ______
• Chat sessions: ______
• Emails: ______
• Social media messages: ______
• Self-service interactions: ______

Total monthly contacts: ______

Step 3: Calculate Overall CPC

Overall CPC = Total monthly costs ÷ Total monthly contacts

Step 4: Calculate CPC by Channel

For channel-specific CPC, you need to allocate costs based on:

• Time spent per channel (from workforce management data)
• Agent utilization per channel
• Technology costs specific to each channel

Simplified channel allocation:

If your agents are channel-dedicated:

Phone CPC = (Phone team costs) ÷ (Phone contacts handled)
Chat CPC = (Chat team costs) ÷ (Chat sessions handled)

If your agents handle multiple channels:

Phone CPC = (Total costs × % time on phone) ÷ (Phone contacts)

Most workforce management platforms provide time allocation data automatically.

Step 5: Compare to Benchmarks

Your overall CPC vs. industry:

• If you're 20%+ below industry average: Investigate quality metrics. Are you cutting corners?
• If you're within ±20% of industry average: You're competitive.
• If you're 20%+ above industry average: Efficiency opportunities exist.

Your channel costs vs. benchmarks:

• Phone: $4.27 - $8.01
• Chat: $2.84 - $4.24
• Email: $3.81
• Social: $3.64
• Self-service: $0.10

If your costs are significantly higher, investigate:

• Agent productivity (contacts per hour)
• Average handle time
• First contact resolution rate
• Channel routing efficiency

Performance Metrics That Drive Cost

Cost per contact is an output metric. It's driven by operational performance metrics.

First Contact Resolution (FCR)

Industry benchmarks:

• Average across all industries: 70%
• Good performance: 70-79%
• World-class performance: 80%+
• Poor performance: Below 65%

FCR by industry:

• Retail: 78% (highest)
• Insurance: 76%
• Health insurance: 72%
• Financial services: 71%
• Energy: 71%
• Tech support: 65%
• Telecommunications: 61% (lowest)

Why FCR matters for cost:

SQM Group research shows: "For every 1% improvement in FCR, you reduce operating cost by 1%."

The mechanism:

• Higher FCR = fewer repeat contacts
• Fewer repeat contacts = lower total contact volume
• Lower volume = lower total costs

The math:

Contact center handling 100,000 contacts/month at 70% FCR:
- 30,000 are repeat contacts (second, third, or fourth attempts)
- Improving to 75% FCR reduces repeat contacts to 25,000
- 5,000 fewer contacts × $5.00 CPC = $25,000/month savings
- Annual savings: $300,000

SQM also found that every 1% FCR improvement drives a 1% improvement in customer satisfaction.

You get both cost reduction and quality improvement from the same operational change.

Average Handle Time (AHT)

Industry benchmarks:

• Overall average: 6 minutes
• Typical range: 4-6 minutes
• E-commerce: 2-5 minutes
• Telecommunications: 8-10 minutes

Why AHT matters for cost:

Lower AHT means each agent can handle more contacts per hour, reducing the number of agents needed.

The math:

Agent works 7.5 productive hours per day (excluding breaks, training)

At 6-minute AHT: 450 minutes ÷ 6 = 75 contacts/day
At 5-minute AHT: 450 minutes ÷ 5 = 90 contacts/day

20% AHT reduction = 20% capacity increase = 20% fewer agents needed

For a 100-agent team with $50,000 fully loaded cost per agent:

• 20% reduction = 20 fewer agents needed
• Annual savings: $1,000,000

The caution: Optimizing for low AHT can backfire.

If agents rush to reduce handle time:

• FCR drops (issues not fully resolved)
• Customer satisfaction drops
• Repeat contacts increase
• Total cost increases

The insight: AHT should be optimized within the constraint of maintaining high FCR and CSAT. The goal is efficient resolution, not fast abandonment.

Service Level (Speed of Answer)

Industry benchmark: 80% of calls answered within 20 seconds (80/20 standard)

Service level doesn't directly drive cost per contact, but it affects staffing requirements.

Higher service level targets require:

• More agents on shift (to handle peak loads)
• Higher overstaffing during slow periods
• Lower agent utilization
• Higher total costs

The trade-off:

• 80/20 service level: 85% agent utilization typical
• 90/30 service level: 75% agent utilization typical
• That's 13% more agents for the same contact volume

Organizations must balance customer experience (fast answer) with cost efficiency.

The Economics of AI and Automation

AI is changing the economics of customer service. But the ROI varies dramatically based on use case and implementation quality, and on whether the AI just answers or actually acts. That distinction is the whole game, and we cover it properly in our agentic AI vs chatbot guide.

AI Support Cost Savings

The headline numbers:

The AI support market will reach $27.29 billion by 2030, growing at 23.3% annually.

95% of customer interactions are expected to be AI-powered by 2025, with leading implementations achieving 148-200% ROI and $300,000+ annual savings.

IBM research shows AI assistants can handle up to 80% of routine inquiries, cutting customer support costs by 30%.

Global cost savings from AI support tools reached $11 billion in 2022 and continue to grow.

The math behind AI support ROI:

In banking and healthcare, each AI-handled query saves approximately 4 minutes of an agent's time, or $0.50 to $0.70 in operational cost per query (Juniper Research).

For a contact center handling 1 million contacts/year:

• 40% automated by AI support = 400,000 contacts
• $0.60 savings per automated contact
• Annual savings: $240,000

The average ROI for AI support tools is approximately 1,275% (and that's just support cost savings, not including revenue impact).

Real-World Case Studies

Klarna (2024):

Leading payment provider Klarna launched an AI-powered assistant in February 2024. In its first month:

• Handled 2.3 million conversations (two-thirds of all customer service chats)
• Equivalent work of 700 full-time agents
• Reduced repetitive inquiries by 25%
• Resolution time dropped from 11 minutes to under 2 minutes
• Estimated $40 million profit improvement in 2024

Alibaba:

During peak shopping seasons:

• AI assistants field 2+ million customer sessions per day (10+ million messages)
• Address 75% of all online customer questions
• Annual savings: ¥1 billion RMB (approximately $150 million)
• Customer satisfaction increased 25%

Vodafone:

Implemented an AI assistant and achieved:

• 70% reduction in cost-per-chat
• Serving customers for less than one-third the previous expense

Telenor:

AI assistant (Telmi) results:

• 20% improvement in customer satisfaction
• 15% increase in revenue
• Significant reduction in human agent workload

The AI Cost Reality Check

These case studies are impressive. But they represent best-case implementations by large organizations with:

• Millions of contacts to amortize development costs
• Dedicated AI engineering teams
• Years of customer data for training
• Sophisticated integration capabilities

For mid-market companies, the economics are different:

AI support platform costs: $5,000 to $15,000/month for AI-powered platforms

To justify this investment:

$10,000/month cost ÷ $5.00 saved per automated contact = 2,000 automated contacts/month needed to break even

At 40% automation rate: 5,000 total contacts/month minimum

The breakeven threshold for mid-market AI support ROI: approximately 5,000 contacts/month.

Below this volume, AI support tools priced per resolution may not beat human agents, which is exactly why the pricing model matters. On a flat monthly rate the breakeven logic changes: the bill doesn't rise with volume, so the more the agent resolves, the lower your effective cost per contact falls.

Where AI Delivers ROI

AI works best for:

• High-volume, low-complexity issues (password resets, order status, FAQ)
• After-hours coverage (AI costs the same 24/7, human agents cost more)
• Initial triage (route complex issues to humans, resolve simple ones)
• Agent assistance (suggest answers, pull up relevant info during calls)

AI doesn't work well for:

• Complex, unique problems requiring human judgment
• Emotionally charged situations requiring empathy
• Multi-step troubleshooting with many variables
• Issues requiring account access or manual intervention

The best ROI comes from hybrid models: AI handles routine work, humans handle complex work, and AI assists humans in real-time.

How an AI Agent Cuts the Cost (in Practice)

The hybrid model above is the right shape, but most AI support tools only do half of it. A chatbot answers the routine question and then hands the actual work, the refund, the address change, the reorder, back to a person. The cost stays in the labour line because the resolution still needs a human.

This is where an AI agent differs from a chatbot, and it's the difference that moves your cost per contact. Omni, our AI agent, reads your live order data through WooCommerce, Shopify and Stripe, so it answers "where's my order?" with the real tracking number rather than pointing at a policy page. And it doesn't stop at answering. When a return is in policy, Omni processes it. When stock is the question, it checks the live figure. The Tier 1 inquiries that make up 40-60% of your inbound get resolved end to end, not just acknowledged.

Three things make that safe to hand over:

• It works to your rules. You set what Omni can resolve on its own, your return window, your refund thresholds, the policies you'd tell a new hire on day one. Omni works inside them, the same way a trained agent would.
• Sensitive actions are queued for approval. Anything above your threshold, a large refund, a goodwill gesture, an edge case, doesn't just happen. Omni prepares it and waits for your yes. You keep the judgement calls; you lose the busywork around them.
• The bill doesn't punish success. Omni is a flat monthly rate, not per resolution or per seat. The more it handles, the lower your effective cost per contact, which is the opposite of how per-resolution pricing behaves. The £250/month founding rate (50% off the £500 plan, for the first 10 places) is on the pricing page.

The result is the hybrid model done properly: Omni clears Tier 1 to your rules, your people keep the cases that need a person, and the repeat work stops landing in anyone's inbox at 11pm. For the longer story of how this works on WooCommerce specifically, see our complete guide to AI for WooCommerce, and if you're weighing tools, our honest comparison of Gorgias, Tidio, Intercom and Omniops lays out where each one fits.

Using Benchmarks for Planning

Benchmarks are useful for planning, budgeting, and evaluating efficiency. Here's how to use them.

Setting Realistic Budgets

When planning customer service budgets:

Step 1: Estimate contact volume

• Historical data (if you have it)
• Industry benchmarks (3-5% of customers contact support monthly typical)
• Seasonality factors (retail peaks in Q4, tax software peaks in Q1)

Step 2: Choose your channel mix

Each channel has different cost and experience characteristics:

Channel Cost Speed Complexity Customer Preference Self-service $0.10 Instant Low only High (for simple issues) Email $3.81 Slow Medium Medium Chat $2.84 Medium Medium High Phone $4.27-$8.01 Fast High High (for complex issues)

Your channel mix should match your customer needs and issue complexity, not just minimize cost.

Step 3: Apply industry CPC benchmarks

Use your industry's average CPC as a starting point:

Budget = (Expected contacts) × (Industry CPC benchmark)

Step 4: Add efficiency assumptions

If you're investing in AI or process improvements:

Adjusted budget = Budget × (1 - Expected efficiency gain %)

Be conservative. Plan for 50% of the efficiency gain you expect. Overestimating AI ROI is common.

Step 5: Add contingency

Customer service volume is hard to predict. Add 15-20% contingency for:

• Unexpected product issues
• Seasonal spikes
• New product launches

Evaluating Make vs. Buy Decisions

Should you run your own contact center or outsource?

In-house economics:

• More control over quality
• Better product knowledge
• Higher fixed costs (facility, management, technology)
• Harder to scale up/down

Outsourced economics:

• Variable cost structure (pay per contact or per agent-hour)
• Easier to scale
• Less control over quality
• Potential knowledge gaps

Outsourcing cost benchmarks:

The global contact center outsourcing market was valued at $97.31 billion in 2024, expected to grow at 9.8% CAGR through 2030.

Typical outsourcing rates:

• Domestic (US-based): $25-$45 per agent-hour
• Nearshore (Latin America): $15-$30 per agent-hour
• Offshore (Philippines, India): $8-$20 per agent-hour

Compare to in-house cost:

In-house agent cost: $50,000 fully loaded ÷ 2,080 hours = $24.04/hour
Add facility, management, technology allocation: $30-$35/hour effective cost

Domestic outsourcing costs about the same as in-house. The benefit is flexibility, not cost savings.

Offshore outsourcing can reduce costs by 40-60%, but comes with risks:

• Time zone challenges
• Language/accent barriers (for some markets)
• Cultural fit issues
• Data security and compliance complexity

The decision framework:

Outsource when:

• Volume is unpredictable
• You lack contact center expertise
• Speed to market matters more than perfect quality
• Cost variability is more important than lowest absolute cost

Keep in-house when:

• Product knowledge is critical and changes frequently
• Customer relationships are high-value and personalized
• You have the volume to justify fixed infrastructure
• Brand experience is a key differentiator

Benchmarking Your Performance

Compare your metrics to industry standards quarterly:

Metric Your Performance Industry Benchmark Gap Cost per contact (overall) $_____ $2.70-$5.60 ___% Cost per phone call $_____ $4.27-$8.01 ___% Cost per chat $_____ $2.84-$4.24 ___% Cost per email $_____ $3.81 ___% First contact resolution ___% 70-79% ___pts Average handle time ___ min 4-6 min ___% Agent utilization ___% 75-85% ___pts Self-service containment ___% 9% (current), 50%+ (goal) ___pts

If you're significantly above benchmarks:

• Investigate root causes (process inefficiency, poor tools, training gaps)
• Prioritize high-impact improvements (usually FCR and self-service)
• Set realistic improvement targets (10-15% annual improvement is good)

If you're significantly below benchmarks:

• Verify your calculations (are you capturing all costs?)
• Check quality metrics (are you cutting corners to reduce cost?)
• Document your advantages (maybe you're genuinely more efficient)

The Limits of Benchmarks

Benchmarks are useful context. They are not targets.

What benchmarks miss:

Customer value: A $100,000 enterprise customer deserves a different service level than a $10/month consumer. Cost per contact should vary by customer value.

Strategic positioning: If premium service is your brand differentiator, you should spend more than benchmarks suggest.

Issue prevention: The cheapest contact is the one that never happens. Investing in product quality, better documentation, or proactive communication may increase short-term costs while reducing long-term contact volume.

Revenue impact: Customer service isn't just a cost center. Support interactions drive retention, upsells, and word-of-mouth. The ROI calculation must include revenue effects, not just cost.

Competitive context: If your competitors provide terrible service, matching industry benchmarks may give you a meaningful advantage. If your competitors excel at service, you may need to exceed benchmarks.

The bottom line: Use benchmarks to understand your relative position and identify opportunities. Don't use benchmarks to set strategy. Strategy comes from customer needs and business goals, not industry averages.

Key Takeaways

Cost per contact varies dramatically by channel:

• Self-service: $0.10 (but only 9% fully resolve)
• Email: $3.81 (often requires multi-message chains)
• Chat: $2.84 (with 3 concurrent conversations)
• Phone: $4.27-$8.01 (one-to-one, immediate)

Industry matters:

• Retail/e-commerce: $2.50-$4.50 per contact
• Telecommunications/tech: $8.00-$12.00 per contact
• Difference driven by issue complexity and regulatory requirements

Labor is the largest cost (60-70% of total):

• Median agent wage: $20.59/hour ($35,600/year)
• Geographic variation: 70% difference between states
• Industry variation: 2-3x premium for technical industries
• Fully loaded cost: 35-50% above base salary

Performance drives cost:

• Every 1% FCR improvement = 1% cost reduction
• Industry average FCR: 70% (world-class: 80%+)
• Average handle time: 4-6 minutes (varies by industry)
• AHT optimization must maintain quality

AI delivers significant ROI at scale:

• Average AI support ROI: 1,275%
• Leading implementations: $300,000+ annual savings
• Breakeven threshold: ~5,000 contacts/month for mid-market
• Best use: routine inquiries, after-hours, agent assistance

Use benchmarks as context, not targets:

• Understand your relative position
• Identify efficiency opportunities
• Make informed build vs. buy decisions
• But strategy comes from customers, not averages

Next Steps

To bring your customer service costs down without cutting quality:

1. Sort your inbound into the three tiers (automate, assist, keep human) and count what share is Tier 1. That number is your headroom.
2. Calculate your actual cost per contact by channel using the methodology above, so you know what a resolved Tier 1 inquiry is worth.
3. Automate Tier 1 with an agent that reads live order data and can act, not a FAQ bot that hands the work back to a person.
4. Raise first contact resolution before you chase marginal handle-time gains. It's the lever that cuts cost and lifts satisfaction together.
5. Reassess quarterly as your business scales and the inquiry mix shifts.

The goal isn't the lowest cost per contact. It's the lowest cost for the same quality, which comes from clearing the repeat work, not rushing the hard cases.

If you'd rather see what that looks like on your store than model it on a spreadsheet, our pricing is one flat rate with every capability included, and the first 10 founding places are £250/month, half the £500 plan.

---

Sources:

• Top contact center statistics & benchmarks (2025) - Plivo
• Cost Per Channel - CallCentreHelper
• Gartner: Only 9% of Customers Report Solving Issues via Self-Service
• Customer Service Representatives - U.S. Bureau of Labor Statistics
• Customer Service Representative Salary - Salary.com
• First Call Resolution Rate Industry Benchmarks - Fullview
• Call Center FCR Benchmark 2024 Results by Industry - SQM Group
• First Call Resolution (FCR): A Comprehensive Guide - SQM Group
• Average Handle Time (AHT) - Zendesk
• What is Average Handle Time (AHT) - GetVoIP
• Chatbot ROI Calculator - Aisera
• Measuring AI Chatbot ROI: Case Studies - Dialzara
• How AI Chatbots Cut Customer Service Costs - NexGen Cloud
• AI Chatbot Statistics 2025 - Fullview
• Chatbot Statistics & Trends 2025 - Tidio