Consumers Aren't Anti-AI; They're Anti-Bad Customer Service, Says Latest Ada Research

If you’re investing in AI for customer service, you’re probably already tracking the usual efficiency metrics: deflection, handle time, cost per contact. But in an interview with CX Today, Dani Wanderer, Chief Marketing Officer at Ada, makes the case that those numbers can hide a much bigger issue. Customers don’t care whether an interaction was handled by AI; they care whether it was actually resolved. And right now, many enterprises are falling short where it matters most. 

Ada’s latest research report, Agentic CX in 2026: What Consumers Expect & Most Enterprises Miss, uncovers a striking disconnect between enterprise expectations and real customer outcomes. The headline stat is hard to ignore, revealing that only 26 percent of consumers achieved full resolution without human intervention in their last service interaction. Not partial answers or a helpful link. Actual completion. 

That gap matters because, as Wanderer puts it, “Consumers aren’t anti-AI, they’re anti-bad customer service.” The problem isn’t that people are refusing automation. It’s that too many deployments are optimized for the business, not for the customer. 

In fact, the data suggests consumers are more open to AI than many leaders assume — with an important condition. Around 59 percent of customers say they’d prefer always-on AI service over waiting for a human… if resolution is guaranteed. The implication is that customers will happily choose speed and convenience, but only when the system can truly deliver. 

That’s where many organizations are getting stuck. Companies may celebrate reduced wait times and lower contact volumes, while customers are still bouncing between channels, repeating themselves, or stuck in limbo. And that mismatch shows up in how organizations prioritize measurement. In the research, consumers rank resolution as the #1 metric, while businesses rank it #7. In other words, many teams are measuring success in ways customers simply don’t recognize as success. 

One of the most practical (and counterintuitive) takeaways from the research is that customers are often willing to trade “human-ness” for competence. Wanderer argues that performance is the real differentiator:  

“Consumers are happily willing to trade a warm conversational empathetic voice for AI that just does the job.”  

That’s a clear signal to CX leaders to stop over-investing in personality layers when core workflows still break. 

Escalation Is a Feature, Not a Failure 

The report also tackles the thorny issue of escalation. Customers want options, and they want them quickly when things go wrong. Ada’s research shows 57 percent of consumers will quit a service if they can’t reach a human, yet only 14 percent of businesses offer frictionless escalation. Wanderer’s framing is direct:  

“Treat escalation… as a feature, not a failure.”  

Rather than viewing handoff as a defeat for automation, it’s a safety net that builds trust, protects retention, and prevents the AI experience from becoming a dead end. 

Trust, of course, is shaped by transparency as well as outcomes. The report finds that 74 percent of consumers expect AI disclosure at the start of an interaction, but 23 percent of businesses either never disclose, or only disclose at the point of handoff. When customers feel surprised or misled, even a technically “correct” experience can be perceived as poor service. 

Governance Is the Unlock for Scaling AI Safely 

Underneath all of this sits a bigger operational challenge: governance. As AI becomes more agentic, moving from answering questions to taking actions, enterprises need clarity on ownership, controls, and accountability. Yet Ada’s research shows that 80 percent of organizations lack an aligned AI governance framework, and only 20 percent have a framework with universal agreement internally. Dani’s view is that governance isn’t what slows teams down; it’s what allows them to scale responsibly.  

“Governance is not really a constraint… It’s really the unlock.” 

So what should CX and service leaders do now, especially if they’re earlier in their AI journey?  

Wanderer recommends a pragmatic approach. Don’t wait for the perfect roadmap; start with one high-volume, repeatable workflow and test it end-to-end, but define success properly. That means measuring true resolution (complete, correct, on-brand, no human needed) and separating performance reporting across AI-only, hybrid, and human-led journeys (a discipline only 28 percent of businesses currently follow). 

The message throughout is urgent but actionable. As Wanderer warns,  

“The cost of waiting to do something is not zero.”  

Expectations are rising, AI is becoming table stakes, and the brands that win won’t be the ones that automate the most, they’ll be the ones that resolve the most. 

Watch the full interview for the complete discussion and practical guidance. For deeper detail, you can also download the full report at Ada.cx