The new economics of customer service: rethinking the cost of growth
When AI makes service more affordable, new demand emerges and more becomes possible. Here's how businesses can rethink growth, investment, and the value of service.
Paulette Chafe
Senior Director, Brand Strategy & Content at Zendesk
Last updated 18 November 2026
In a recent analysis of a large cohort of Zendesk AI customers, we found a striking pattern. At a median automation rate of 41%, total service interactions had roughly doubled. Yet agent headcount had increased by just 3.5%.
Businesses were handling substantially more service activity without expanding their teams at the same rate. Zendesk CTO Adrian McDermott points to Jevons Paradox, an economic principle that helps explain this surprising relationship. When technology makes something more efficient and affordable, demand for it can actually increase.
For customer service, this raises a bigger question: what happens when businesses can afford to meet customer needs they previously couldn't serve?
The answer has implications for how businesses grow, where they invest, and what they expect service to accomplish.
Greater efficiency can stimulate demand, making it economically viable to serve needs that previously went unmet.
Greater efficiency makes more service possible.
For years, businesses have approached AI in customer service primarily as an efficiency opportunity. Automate more requests, reduce the cost per interaction, and increase productivity without adding headcount.
Those benefits are real. But they assume the amount of service customers need is relatively fixed. In practice, demand for service is influenced by how easy it is to access and how much it costs to deliver.
Customers don't contact a business every time they encounter a problem. They weigh up whether the issue is worth the effort of finding help, explaining what happened, and waiting for a resolution. Some abandon the process. Others put up with the inconvenience or find another solution.
Businesses make similar calculations. They prioritize the requests they can afford to handle, design support around available resources, and accept that some customer needs will go unmet. When AI lowers the cost of delivering service, those constraints begin to change.
Businesses can make help available in more situations, address requests that were previously too expensive to serve, and offer assistance that once required significant human effort. Customers have fewer reasons to delay or avoid asking for help. Greater efficiency can stimulate demand, making it economically viable to serve needs that previously went unmet.
Consider Arajet, the Dominican Republic-based airline. As passenger volume grew approximately 40 to 50% year over year, the company used Zendesk to unify its service channels, introduce AI agents to handle routine inquiries, and equip employees with Copilot. The results were a 40% reduction in cost per interaction, 60% more tickets resolved per agent, and stable customer service costs despite rapid growth.
As Zendesk CTO Adrian McDermott puts it: "The interesting question isn't how many people we can cut, but how much better service can become when the next interaction no longer carries the same cost."
AI is changing the economics of getting work done
The cost of a service interaction is only one part of the cost of solving a customer problem. A request might take minutes to acknowledge but require hours of work behind the scenes. Employees need to gather information, check policies, move between systems, secure approvals, and coordinate with colleagues before the customer's problem can be resolved.
Consider an employee requesting access to a business application. The initial request is straightforward, but fulfilling it might involve checking eligibility, obtaining approval, updating an identity management system, and confirming access. The employee submits one ticket, yet completing the work requires coordination across several systems and people.
AI agents can increasingly take on that coordination, retrieving relevant policies, initiating workflows, carrying out approved actions, and involving human specialists where judgment is required.
We're already seeing the scale of this opportunity. In their first two months of customer use, our Employee Service AI agents executed 9 million workflows across 300 customers, representing work we estimate would otherwise have required 750,000 hours of manual effort. The potential savings extend beyond the time spent handling requests. They include the manual work, handoffs, and delays involved in completing them.
There's another economic opportunity in addressing the problems that generate those requests. Businesses can become highly efficient at resolving recurring issues while continuing to incur the cost of the underlying problem. Repeated employee access requests, for example, might reveal a poorly designed onboarding process.
AI can help identify those patterns, investigate where workflows break down, and support improvements that prevent the same problems from recurring. That creates two opportunities for businesses: reducing the effort required to complete work and reducing the amount of avoidable work that needs to happen in the first place.
The next investment opportunity is in outcomes
These changes raise an important question for service leaders: where will the next investment in AI create the greatest value? Traditional service measures provide part of the answer. Cost per contact, resolution time, ticket volume, and automation rates help businesses understand operational performance. But they don't always capture the value of the work being completed.
A business could automate thousands of routine inquiries and achieve substantial savings. Another investment might address a smaller number of complex issues but prevent expensive operational delays, reduce repeat contacts, or protect valuable customer relationships.
Evaluating those opportunities requires a broader understanding of what service contributes to the business. A completed interaction isn't necessarily the same as a successful outcome. A technical support ticket might be closed even though the underlying problem returns a week later. A refund might be processed successfully while the customer relationship is still lost.
Customers recognize the difference. Our latest research shows that 70% of consumers say AI has made service faster but not necessarily better. And 81% would rather wait for a genuinely resolved issue than receive an instant response that doesn't solve their problem.
For businesses, this means evaluating operational efficiency alongside the results service creates. That might include the cost of completing an end-to-end workflow, the number of repeat issues prevented, the time saved across departments, or the financial impact of resolving a business-critical problem.
Docebo, the enterprise learning platform, offers an example of how these considerations can influence investment decisions.
The company had grown rapidly, but its support organization was working across ten disconnected tools. Employees spent time navigating systems, gathering customer information, and piecing together reports, making it difficult to identify service trends or take a more proactive approach.
Docebo consolidated its support operations on Zendesk and introduced AI agents, Copilot, and AI-powered quality assurance. AI agents now resolve 43% of incoming chats and 30% of emails, freeing specialists to concentrate on more complex customer issues. Copilot reduces the time employees spend writing responses and reviewing long customer conversations, while AI-powered QA evaluates 100% of interactions and helps the business identify recurring problems and potential churn risks.
These changes reduced Docebo's cost per ticket by 52% and saved $125,000 in technology-stack costs. The company also estimates nearly $149,000 in annual efficiency savings from its increased use of AI. The significance goes beyond the savings themselves. Docebo has created greater visibility into service quality and customer needs, giving its teams more information to identify problems and improve how support is delivered.
This broader shift toward outcomes is also shaping Zendesk's approach to AI. With the recent introduction of the Zendesk Outcome AI™ Platform, we're moving beyond helping businesses manage and resolve interactions to helping them complete the work that delivers meaningful results.
Powered by the Autonomous Service Workforce, the platform brings together specialized AI agents, people, knowledge, and connected systems to coordinate more complex work. It also helps businesses learn from outcomes, identify recurring problems, and continuously improve how service is delivered.
Investing in what service can accomplish
AI is changing what businesses can afford to deliver. Lower costs make more customer needs economical to serve, while more capable AI expands the work businesses can complete.
For service leaders, this creates an opportunity to rethink where they invest: which customer needs can now be addressed, which processes are worth redesigning, and where better outcomes can create the greatest business value.
The businesses that benefit most will use AI's efficiency gains to expand what service can accomplish, with a clear understanding of the value those investments deliver.
Paulette Chafe
Senior Director, Brand Strategy & Content at Zendesk
Paulette Chafe is Senior Director, Brand Strategy & Content at Zendesk, where she shapes the company’s brand, customer insights, thought leadership, and executive content—helping drive brand reputation and pipeline growth. Her team identifies emerging developments and research-based insights relevant to customer experience, service and IT leaders, including through Zendesk’s annual flagship report, CX Trends.
Paulette is known for building high-performing teams and tackling complex challenges with focus and clarity. Passionate about technology, especially AI, she explores how it can amplify human work, make strategies smarter and more efficient, and help shape the future of contact center work.
Make more outcomes possible with AI
Zendesk's Outcome AI™ Platform brings together specialized AI, people, knowledge, and connected systems to help businesses complete work, solve problems, and continuously improve service.
Zendesk's Outcome AI™ Platform brings together specialized AI, people, knowledge, and connected systems to help businesses complete work, solve problems, and continuously improve service.