Voice Is the Last Line of Defense in CX. Time to Treat It That Way.

When self-service can't resolve it, when the app doesn't have an answer, when the stakes are too high to rely on a chatbot, customer care all comes down to the voice channel.

Voice is the last line of defense in customer experience (CX). Yet many organizations still approach it as a component of the legacy contact center, treating it with a "good enough" mindset while newer digital initiatives get the focus and investment. As that happens, voice keeps playing host to some of the most persistent legacy contact center challenges: phone trees that go in circles, chatbots that lack understanding, and live agents that require repeated explanations.

Voice carries outsized influence over CX and cost, but it keeps losing out on budget and attention to initiatives that are easier to point to on a roadmap. That’s why the current state of voice is so hard to fix.

That's what I want to lay out here: what voice actually looks like right now, why it keeps falling short of what customers need despite how much they rely on it, and what it will take—including the AI-assisted solutions organizations are increasingly turning to—to close that distance for good.

The Voice Channel: Customers Haven't Stopped Choosing It

Voice isn't declining. And it’s not a low-stakes channel. Three in five contact center leaders admit they've tried, and failed, to shift customers off the phone and onto cheaper, more convenient channels.

The persistent reality is that customers call when they need help, and they want a more definite answer or resolution than they feel they can get from digital channels or self-service. They resoundingly identify the phone as the most trusted support option they have, ahead of every digital channel on offer. And it's not just an old habit reserved for older generations: 71% of Gen Z and 81% of Millennials said they're likely to call the contact center to resolve an issue when they've already attempted self-service, McKinsey found.

And for a lot of customers, voice isn't a preference at all: It's the only door that works. For someone with a visual impairment, a motor limitation, or a cognitive disability, navigating a web form or a chat widget ranges from difficult to impossible. Voice is often the most reliable way these customers get help, and sometimes the only way, so a broken IVR shuts them out entirely.

Voice is the channel customers turn to specifically when an issue is complex or emotionally loaded. Misroute a billing dispute, and it doesn't just frustrate the customer; it can escalate into a real collections issue. Digital self-service is great for a balance check, but it's not what people reach for when the issue becomes confusing or urgent.

Customers deliberately choose voice, again and again, for the moments when a good experience matters most.

"How did the industry get to this point, where voice remains pivotal yet deprioritized? The answer is how the technology has developed."

— Gabriel Sjoberg, Director of Software Development, CSG

Why Voice Carries a Different Kind of Risk

Generally speaking, failures in the voice channel feel worse to customers than failures in digital. A clunky app screen or a slow-loading page is an annoyance; a customer can retry, refresh, or just come back later with no real record that anything went wrong.

But there's no undo button on a phone call. If the system mishears an account number or forces someone to repeat themselves for the third time, that moment is final the instant it happens.

Even if customers keep opting for voice, hoping it gives them fast answers, it doesn't necessarily mean they expect it will. At least when it comes to the automated system that greets them: a recent CMP survey found that only 14% of consumers are confident an IVR can resolve their issue. That's a lot of customers dialing in with their finger already hovering over the zero.

That lack of trust reflects the state of the traditional IVR customer experience, which many callers still associate with rigid menus and limited understanding of intent.

Customers are very ready to shove past an AI chatbot, too, when it answers their call. PwC found that roughly half of consumers demand a direct connection to a live agent the moment a chatbot fails to answer their question. And we know in that scenario, organizations struggle to route the caller to the best agent group to handle the inquiry.

So to sum up, customers are still choosing voice, but

1. It's often because the self-service options didn't/don't work

AND

2. They don't often have much faith that it’ll resolve their issue

So when voice fails, that right there constitutes multiple bad experiences—at which point nearly 53% of consumers are willing to switch to a competitor outright, according to a CCW Digital Market Study.

This is why viewing voice solely as a legacy contact center function is hiding a real cost to the enterprise. Every dollar and every engineering hour that voice doesn't get—on the theory that it's a mature, low-priority piece of the stack—is a dollar/hour not spent on the channel customers often distrust yet must still rely on.

RELATED ARTICLE: CCW Las Vegas 2026: The AI Conversation Is Growing Up 

Forty Years of Answering the Phone: The Evolution of the IVR Customer Experience

How did the industry get to this point, where voice remains pivotal yet deprioritized? The answer is how voice technology has developed.

The IVR is roughly 40 years old. For its time, it was a genuine leap forward. It let a single phone line sort incoming calls into queues without a human touching every one. That single innovation is the origin of the traditional IVR customer experience most callers still encounter today. But what IVR never did, in four decades of iteration, is understand the person calling. It could route you. It couldn't listen to you.

That's what I'll spend the next post unpacking: It's the detail that explains why voice AI has been so hard to get right: for 40 years, this technology has been very good at understanding words but not very good at understanding people. Everything that's changed in recent years—the shift from IVR to conversational AI to agentic AI—is really an attempt to close that specific, stubborn gap. And everything that still goes wrong in production traces back to how hard that gap is to close for real, at scale, on a live phone line, with a real customer on the other end. But today we can do it.

Next:

We'll look at why a voice AI system can look flawless in a demo room and still fall apart the moment it meets a real caller. And I'll talk about ways organizations can address that.