Most AI customer service deployments have a resolution problem. New research from Ada and NewtonX reveals why businesses can’t see it.

Ada, the leader in agentic customer experience, and NewtonX, the world’s leading B2B market research firm, announced groundbreaking new research on AI in customer experience. Agentic CX in 2026: What consumers expect and most enterprises miss is a dual-sided primary research study that helps answer the biggest question enterprise CX teams are wrestling with right now: Is AI in customer service actually working?

Consumers are more open to AI-powered customer service than many assume—59% prefer instant, 24/7 AI support over waiting for a human. But that openness hinges on one condition: AI has to actually resolve their issue. That resolution consistently falls short for most businesses, the research shows, and few have the operational infrastructure to even realize it. 

Where the gap lives

The study reveals a consistent pattern across industries and geographies: businesses are scaling AI investment while optimizing for the wrong outcomes. Consumers want resolution. Businesses are measuring deflection. 

“Most businesses are still optimizing for the most familiar metrics: deflection, containment, and cost savings. But those are now table stakes,” said Mike Murchinson, CEO of Ada. 

“What this new data reinforces is what these metrics don’t tell us: was the customer’s problem actually solved? CX leaders are now being asked to demonstrate that AI drives business outcomes beyond just efficiency and cost-cutting.”

That misalignment shows up across several key findings in the data: 

• The resolution gap: Only 24% of consumers had their issue fully resolved by AI in their most recent interaction, and 11% abandoned entirely with no complaint filed or follow-up. Quiet exits that most businesses have no system to catch.

• The empathy myth: Consumers ranked the importance of accuracy and problem-solving nearly three times higher than empathy. Given a choice between a neutral AI resolving a billing issue in 10 seconds and a conversational AI taking 2 minutes with small talk, 59% chose speed. Warmth is nice, but resolution is the job.
• The measurement blind spot: Resolution rate ranks ninth among the benefits businesses track from AI deployments. Reduced wait times ranks first. The metrics most businesses optimize for have almost no connection to what consumers say they actually need.
• Ambition ahead of readiness: The share of businesses operating with primarily AI-led CX is projected to grow 2.4x over the next 12 months. Yet 80% lack a fully adopted governance framework for AI in customer service, meaning most are planning to scale an operation they don’t yet have the controls to run safely.

“Our research shows that many companies are scaling AI without a clear view of how it’s affecting the customer experience. When businesses measure AI and human performance together, it becomes much harder to see where the technology is breaking down,” said Leon Mishkis, COO at NewtonX. 

“If an interaction fails because the system doesn’t understand the customer, that’s not just a technical problem. It’s a sign the business is losing touch with what customers need.  In 2026, always-on service is only an advantage if it’s reliable. If it isn’t, AI becomes a much faster way to damage brand trust.”

How the leaders are pulling ahead

Maturity in agentic CX doesn’t happen in a single deployment decision. The full report lays out a four-stage model showing where most enterprises are currently stuck, what it takes to advance, and what becomes possible at each stage, from FAQ automation to fully AI-led operations with real-time personalization. The Ada ACX Blueprint, included in the report available March 26, gives CX leaders a concrete framework to assess where they are and what to build next.