Just 4% of Service Leaders Hit Their Expected AI Returns, Gartner Finds

Only 4% of customer service organizations have achieved the AI ROI they expected, Gartner told attendees at its Customer Service & Support Conference in Aurora, Colo., on Nov. 4-5. The number that should worry CMOs more: 76% can't show any positive return at all, even as AI spending among service leaders jumped 38% while overall service budgets grew just 2%.

The ROI gap explains why rehiring has started. A Gartner survey of 275 service and support leaders, run in July and August, found only 21% had cut human agent headcount because of AI. Among those that cut, the average reduction was 16.7% of the workforce over 18 months. And 7% have already rehired workers after earlier reductions, with another 11% planning to rehire within 18 months. Plans ran ahead of action: in April, 31% of service leaders had implemented or planned AI-driven frontline layoffs.

Christopher Sladdin, senior director analyst at Gartner, said the low ROI figure was expected. The surprise was elsewhere.

"Given the rising costs for AI … and the challenge in converting time savings into financial savings, we aren't surprised that so few organizations have realized the full ROI," he said. "But what is more surprising is how only 24% of service organizations report seeing any positive ROI on their investments."

If three-quarters of organizations can't show a positive return, cuts justified by that return were made on projections. That's how rehiring starts. In February, Gartner predicted half of companies that cut service staff because of AI would rehire by 2027, often under new job titles. The 7% figure from Aurora is the first sign that prediction is tracking.

Why didn't AI replace service agents?

Daniel O'Sullivan, senior director analyst at Gartner, described the original expectation plainly: in 2023, customer service was supposed to be "the first domino in the enterprise to fall" with "massive reduction in headcount." Several years on, he said, organizations "haven't necessarily reached the pot of gold at the end of the rainbow."

Executive pressure was real — 91% of service leaders faced it in 2026. The results were modest. Channel strategy explains part of the shortfall, according to Emily Potosky, senior director analyst at Gartner.

"We've been kind of mean to the phone. And we've treated the phone channel as a must-have and a fallback, but something where we really prefer if our customers emailed us or chatted us," she said. "Regardless of the investments that most leaders have made, customers still call. One third of customers use the phone on a customer service interaction. The only channel that beats that is website."

Phone demand stayed when the agents left. That goes a long way toward explaining the rehiring.

For marketers, the stakes go beyond staffing. Human agents handle escalations, complaints and retention saves — the interactions that feed voice-of-customer programs, CSAT dashboards (customer satisfaction scores) and churn models. Cuts that outrun AI readiness thin that signal, and customers absorb the difference through longer resolutions and repeat contacts. Voice also carries tone, urgency and intent that text channels miss.

What separates the 4% that hit their targets?

Sladdin's opening keynote identified three shared habits among the winners: they use AI to learn from customer interactions at scale; they build connected portfolios of "fewer, higher-value and interconnected use cases"; and they invest nearly 1.8 times more in data and knowledge management foundations than in the AI use cases themselves. That 1.8x ratio is a budgeting benchmark marketing can use, since knowledge content and customer data are typically shared across both functions.

Kathy Ross, VP analyst at Gartner, warned of the "cut fast, rehire later" trap that appears "when they reduce headcount based on expected AI value rather than demonstrated results." Rehiring means paying again to recruit, onboard and train, and rebuilding expertise that already walked out the door.

What is the 'AI customer' and why does it need two tracks?

Sladdin flagged a newer development: customers now send AI agents such as Meta Muse and OpenAI Dots to handle service on their behalf. "In just a few months, we've seen rapid adoption of these technologies by end customers, and service and support leaders are increasingly observing customers' AI agents working with their human customer service representatives," he said.

His answer is a "two-track CX" — offering both an AI-led and a human-led path to every customer outcome, rather than treating humans as a fallback. O'Sullivan cautioned on Day 2 that lower customer effort can raise service demand while weakening direct brand relationships, and urged leaders to build an "agent-operable enterprise."

Potosky extended the idea to voice beyond phone — search and in-app voice interactions — a pattern marketers tracking brand visibility in AI answers will recognize.

What should CMOs do before the next cut?

Gartner's own numbers suggest caution. Roughly as many service organizations report negative AI ROI as positive, Sladdin noted, with an even larger share "unaware of what they're realizing from these investments." Practical steps:

• Set measurement before workforce changes: baseline CSAT, repeat contacts and resolution time first.
• Ask vendors for live-deployment evidence, including Gartner's five trust tests — disclosure, choice, explainability, data boundaries, recourse.
• Check the foundation budget against the 1.8x ratio.
• Compare phone's share of contacts against the one-third benchmark.
• Audit returns, billing and cancellation for both an AI-led and a designed human-led path.

Gartner's 2027 rehiring prediction gets tested this year. The 7% who already rehired suggest the measurement, not the promise, should drive the org chart.