Why Bad AI Is Costing You Customers in 2026

New Verint research shows poor self-service is driving dissatisfaction – and what fixing it actually looks like

4
Sponsored Post
Why Bad AI Is Costing You Customers in 2026
Contact Center & Omnichannel​Interview

Published: May 12, 2026

Rhys Fisher

More than half of customers now say businesses fail them when they need help. At the same time, 42% report their expectations have risen in 2026, compared to 36% in 2025.  

Two numbers moving in opposite directions, and the distance between them is getting larger.  

These are two of the headline statistics from Verint’s State of Customer Experience 2026 report 

According to Kelly Koelliker, VP of Product Marketing at Verint, the reason behind these worrying findings is that budget pressure has forced contact centers to do more with less, while the rush to plug that gap with AI self-service has, in many cases, backfired.  

Koelliker explain:

“A lot of the contact centers we talk to are saying: I need more people, but they won’t let me hire more people.” 

“So when that happens, your customer experience is going to suffer. A lot of organizations have tried to put some AI on their website to avoid answering questions, but they’re not putting the right thought and process into how they’re doing it. It’s causing a lot of bad CX.” 

The Boiling Point  

Nowhere is that frustration more visible than in the report’s human preference data.  

61% of customers now prefer speaking to a human agent over an AI-powered service – up five percent year-on-year.  

More striking is the 18% rise in human agent preference among 18–34-year-olds: the demographic most brands have counted on as their natural self-service audience.  

Koelliker, however, is not surprised. To illustrate the point, she details a real-life example from her local Facebook community group, where a customer returned a perfectly functional product, not because of any fault with the item, but because they refused to deal with the company’s customer service experience.  

“People are tired of just yelling ‘representative’ into the phone,” she says. “It has reached a boiling point.  

“People have hated bad self-service for a long time, but now they’re willing to take their business elsewhere. That’s how bad it’s gotten.”  

But this isn’t a ‘human = good, AI = bad’ story. Indeed, 69% of customers who currently prefer human agents say they would switch to AI if it could fully resolve their issue.  

They are not opposed to automation; they are opposed to automation that fails them.  

Koelliker says:

“Buyers feel offended. You can’t just look at efficiency and say we win. You have to say: ‘My customer should leave this interaction thrilled.’ Otherwise, it’s not worth it.” 

Getting AI Self-Service Right  

Koelliker describes what good AI self-service requires as a four prong” framework:  

  1. Generative AI: Provides the base by surfacing answers from company content.  
  2. Intent Matching: Layered on top, particularly effective in regulated sectors where specific, compliant answers are required.  
  3. Agentic Capability: Completing tasks, not just answering questions.  
  4. Smart Transfers: Knowing when to transfer to a human, doing it before the customer demands it, and not making them re-explain their request.  

“People don’t want to just ask a question. They want to actually do a thing: fill out an application, make a payment, change a shipping address,” she says.  

“If your chatbot just says, ‘Oh, if you want to do that, call us,’ that’s where people get very frustrated.”  

For organizations not ready to rebuild their self-service stack, Koelliker points to a faster win: using AI within existing agent-assisted interactions.  

Verint’s companion State of Agent Experience 2026 report, covered separately by CX Today, found that agent workflows are packed with manual tasks AI can absorb.  

“We’ve seen people double or triple agent capacity by taking out all that busy work,” Koelliker says.  

Phone Is Not Done Yet  

The report also captures a fairly significant channel shift, highlighting that digital preference has plateaued at 70%, down from a 2025 peak of 73%, while phone preference has risen for the first time in three years to 30%.  

66% of customers contacted a company by phone in the past twelve months – more than any other channel type.  

“There was a period where digital kept going up, and everyone started to think they could ignore their phone,” Koelliker says.  

“It has to be both – and not just both separately. It has to be a unified experience across all of it, because customers are switching between channels within a single interaction.”  

CX Is a Revenue Number  

The report’s final finding is one the industry keeps being reminded of, and keeps struggling to act on.  

Exceptional CX ranks as the second most important driver of customer loyalty, just four points behind product quality.  

80% of customers would repurchase and recommend after a great experience, whereas 79% would switch to a competitor after a bad one.   

Koelliker’s concern is that the businesses most likely to act on these numbers are already the ones invested in CX measurement: 

“The companies that aren’t caring, they’re not measuring, they’re not fixing it. They’re going to keep going downhill, and they’re going to lose their business to brands that may even have an inferior product but a better experience.” 

Speed is what customers want most, the research confirms. In 2026, it also serves as the baseline. The brands closing the experience gap are the ones delivering it without asking customers to choose between fast and good.  

Agentic AIAgentic AI in Customer Service​AI AgentsAutonomous AgentsCustomer Retention StrategiesDigital Customer Service PlatformsUser Experience
Featured

Share This Post