Quick Overview

  • According to an analysis of 432 use cases released by Gartner last month, only a quarter of AI customer service use cases generate a return on investment.
  • Another quarter of use cases yield negative returns, and 42% have unclear returns—support leaders say they cannot determine the value created. Only 11% of customer service use cases break even.
  • Despite unclear returns, more than three-quarters of leaders plan to increase AI investment in 2026.

Deep Insights

Although executives at large companies from Verizon to Airbnb tout the success of their AI chatbots, for most companies, cost savings from AI investment in customer service remain elusive.

Among business functions, customer service is a leading area for enterprise AI adoption. Gartner found that customer service and support teams advance an average of nearly 5 AI use cases and allocate about 13% of their functional budget to AI.

However, customer service leaders are increasingly accountable for a metric they often cannot prove—more than half (56%) of service and support leaders expect their incentives to be directly tied to AI outcomes in 2026.

Experts say this disconnect stems from a top-down approach: it fails to address actual customer needs and oversimplifies assumptions about resolution rates and staffing.

“What we see in deployments is that everyone is trying to bring in AI,” Antoine Nasr, AI lead at Forethought AI Agents, part of Zendesk, told CX Dive. “It's a top-down initiative: we need AI, we need customer service and customer experience.”

Often, when leadership directs customer service departments to implement AI, it does not start from a clear customer problem.

“Too many AI deployments start with the pressure to present a credible AI strategy to the board, rather than starting with a well-defined business problem,” Julie Geller, principal research director at Info-Tech Research Group, told CX Dive in an email.

Many enterprises expect to achieve cost savings by reducing headcount, letting AI agents take over many simple issues handled by customer service representatives. But according to Gartner, the proportion of organizations adding staff is comparable to those cutting staff: about a quarter of companies report headcount growth, and about a quarter report reductions.

As more companies adopt AI, they also need to hire new specialized roles to manage AI.

Using resolution rate as a goal is also misleading if it does not help customers.

“Resolution rate is also often mistaken for success,” she said. “Delaying transfer to a human agent does not mean the customer's problem is solved. The real test is much simpler: did the customer get what they needed with less effort?”

Gartner's research aligns with a recent report from Forethought AI Agents, part of Zendesk. Although 70% of organizations have deployed AI in customer experience, only a small fraction are generating value in improving outcomes and return on investment.