Aligning the Gap Between Business Assumptions and Consumer Preferences

Although consumers are becoming more positive about interacting with businesses using AI agents, business leaders think their customers want to use the technology more than they actually do. With adoption, capabilities, and preferences changing rapidly, it’s important for business leaders to narrow the gap between their expectations and reality.

For example, IT/CX/AI leaders say 42.7% of their customers prefer interacting with AI agents and 10.3% avoid them. In reality, only 21.7% of consumers say they prefer using AI agents and 32.7% avoid them, according to Metrigy’s AI’s Role in Customer Experience 2026-27 global study of 759 companies and Metrigy’s 2Q AI Consumer Experience Index, a study in partnership with NiCE, with 1,000 U.S. consumers. 

That does not mean consumers are always avoiding AI agents–or even that they’re having bad experiences. Overall, the Metrigy AI Consumer Experience Index increased quarter-over-quarter, and the AI Satisfaction sub-index saw the largest increase, of 11.8%. The tides are turning, albeit slowly, so business leaders have a balancing act to follow: Don’t roll out AI agents too aggressively without the option to escalate to humans, and don’t be too far behind to frustrate customers who know AI agents’ value.   Business enthusiasm for AI isn’t a sound CX strategy, and assuming customer comfort is a fast way to erode the trust you’re trying to build.

Companies Think AI Can Stretch Further—Customers Aren’t So Sure

The gap widens when we ask companies how far AI can go. Nearly 73% of business leaders say AI agents are capable of doing more than customers are comfortable with them handling, and that figure rises to 81.2% among Metrigy’s research success group (the companies achieving above-average business improvements from AI).

The most successful companies aren’t claiming customers are holding them back out of ignorance. They’re acknowledging a comfort ceiling, and that’s the right instinct. Companies that push past that ceiling—automating interactions customers would rather not hand to AI—don’t save money in the long run. They generate callbacks, complaints, and churn.

Why does the perception gap exist at all? A few reasons:

• Businesses may not have the right analytics in place to judge customer success or sentiment.
• Business analysts are misinterpreting data. Just because an AI agent successfully resolves a high volume of contacts does not mean customers are happy they are there. A customer who gets an answer from an AI agent may have gotten what they needed and still would have picked a person if you’d asked. 
• Consumers may have bias. If they’re thinking of today’s AI agents as yesterday’s v1.0 chatbots, they would be justified avoiding them. They need to know what the latest agents are capable of handling–and that takes marketing.
• Businesses aren’t marketing AI to their customers. For example, if a customer has a quick resolution with AI, why not close out with: “Glad our AI agent was able to solve your problem today in 3 minutes, 5 seconds. If you had waited for a human agent, the estimated resolution time would have been 10 minutes, 15 seconds based on current hold times.” This way, the consumer can internalize the value received for the next interaction.

Where AI Actually Wins, and Where It Doesn’t

So where’s the line? Our research participants are fairly clear-eyed about it. CX leaders say AI agents are more effective than humans for most routine issues. But humans still win for refunds, complaints, and selling products or services.

That split makes sense. Refunds and complaints are emotional and often require judgment or an exception to policy. Selling requires reading a person and adjusting on the fly. Those are human strengths, and they happen to be the interactions where getting it wrong costs the most. AI is the way to go for high-volume, low-emotion contacts. It is not the way to go for the moments that decide whether a customer stays–at least for now. AI has the technical capability to handle some of these more complex issues, but businesses first must earn their customers’ trust in the use of AI. For customer-facing AI, take it a little slower and be more methodical.

Consumers tell us the same thing when we ask what they’ll accept on the phone. Nearly 60% are comfortable with accepting a call from an AI voice agent for issues such as appointment reminders, delivery status, fraud alerts, billing and payment reminders, and service renewals. Notice the pattern: these are informational and transactional. None of them ask the customer to negotiate or vent. Build your voice AI around that list, and you’re working with customer preference instead of against it.

AI Triage: The Front Door Most Customers Now Walk Through

Credit where it’s due: companies are getting triage right. Two-thirds (66%) now open at least some customer interactions with an AI triage agent. On average, 36.3% of customer interactions start with an AI triage agent, which replace the old touch-tone IVR, letting callers say what they need in plain language instead of mashing numbers into a menu.

That’s a smart place to lead with AI, and the spending backs it up: 56.1% of companies increased their investment in AI triage in 2025. The catch is cost. Triage runs more expensive than IVR, which is why some CX leaders use a hybrid—IVR to offload high-volume or low-priority calls, triage for everything else. That’s a reasonable way to manage the bill without forcing customers back into a menu tree.

Why the Gap Matters

Step back, and the strategic risk comes into focus. If you believe 42.7% of your customers want AI when the real number is 21.7%, you’ll over-invest in customer-facing automation and under-invest in the AI that makes your human agents better. And the human-facing AI—agent assist, real-time coaching, post-interaction summarization— lifts performance on exactly the emotional, high-stakes contacts customers don’t want automated in the first place.

The successful companies already understand this. In fact, 56.7% prioritize AI that helps human agents over pure self-service automation. They’re not choosing people or technology. They’re using technology to sharpen human performance on the interactions that matter, and using AI agents to clear the routine volume so those people have the time to do it.

This explains, in part, why the Metrigy AI Consumer Experience Index increased measurably from first to second quarter this year. Whether customer sentiment is high with either an AI or human agent, AI itself is helping to improve the experience. And the more companies embed AI in their entire CX platform (self-service, agent assist, workforce enablement management, analytics, etc.), the more the flywheel improvement continues. 

What CX Leaders Should Do

Don’t let your own enthusiasm for AI cloud your judgment on what your customers really want:

• Measure the gap yourself. Ask your customers directly which interactions they’d rather handle with AI, and compare that to what your team assumes.  Use interaction analytics to gather that same information in a different and even more objective way.
• Point AI agents at the transactional, high-volume contacts customers already accept: reminders, status updates, fraud alerts, and triage.
• Keep humans on the emotional, high-stakes work, and invest in the AI that assists them rather than the AI that replaces them.
• Run triage as your front door, and consider a hybrid with IVR to control cost.
• Revisit these assumptions every couple of quarters. Customer comfort is rising, but it’s rising at the customer’s pace, not yours.

The companies that close this gap may not be automating customer service the fastest, but they will be the ones that align best with their customer preferences.

Want to more information? Download our complimentary report.

Author(s)

Robin Gareiss