The IVR has been dying for a decade. Why is it still alive?

Somewhere right now, a customer is pressing 1 for unbearable boredom, 2 for elevator music, and 3 to hang up and give up. This isn't a niche edge case or a legacy throwback. It’s the default customer experience most enterprise customers face today.

Despite tech publications writing the IVR's obituary for years, Parloa’s recent research found that 96% of dedicated support lines continue to rely on legacy interactive voice response (IVR) architecture for their customer experience.

Why?

Because companies treat their contact centers as cost centers, and in doing so, settle for good enough… just like anyone would if they assumed little return on their investment. The problem is, by settling for good enough in the cost center, companies erode the customer loyalty go-to-market teams worked so hard to build.

A "good enough" approach to customer experience

The global IVR market is valued at ~$4.6 billion, with analyst projections pointing to 5% annual growth through 2033. The market is mature, and it's steady, but it's no longer accelerating. These market trends highlight the exact reason why the system hasn't died: enterprises have deemed it "good enough." A good enough approach doesn't accelerate, but it's dependable, and when you're investing in something you don't want to spend money on, dependable is good enough.

The modest growth the market does see comes from the little investment enterprises have made in an attempt to look like they’re modernizing systems and delivering better customer experiences with GPT technology.

▶️ Also read: The 5 Voice AI Trends Defining 2026

Why wrapping IVRs in GPT isn't enough

The logic is sensible: rather than ripping out legacy infrastructure and overhauling workflows, simply layer generative AI on top. Through natural language understanding, greet the human in a friendly manner, interpret the customer's initial intent, and then route that intent to a menu option.

While sensible, the logic isn't rational. Wrapping IVR in GPT is like putting a fresh coat of paint on a house. The outside looks modern, while underneath, the house has the same wood panels as before. The GPT "paint" does about as much as putting paint on a house with rotting wood. It's a temporary fix to a much bigger, underlying problem.

Designed to deflect

At their core, IVR systems have been designed to deflect customer contact volume. Operating through what appears to be a self-serve model, companies implement IVR systems to enable customers to self-serve their problems, hypothetically lowering cost-per-contact and improving wait times.

Parloa’s research found that these hypothetical fixes aren't fixes at all. Contact systems are still leaving customers on hold for up to 90 minutes, and humans are being sent down eight menu paths just to try to reach a resolution. While these experiences don't show up as budget line items, the moments of dead air and button presses are the moments where frustration builds and customer loyalty quietly erodes. Qualtrics XM Institute estimates that poor customer experiences will put $3 trillion worth of sales at risk this year, with 34% of bad experiences leading customers to spend less, and 13% to stop spending altogether. The hidden cost of treating a contact center as a cost center ultimately compounds.

The enterprises taking a deflect approach to customer support are making a bet that customers will tolerate friction indefinitely. For a while, this was true. Today, however, customers know the technology is available for better customer experiences, and they’ll go to the companies who deliver it.

▶️ Also read: What it takes to build scalable AI voice agents

What "killing the IVR" actually requires

Killing the IVR requires not just a swap in products, but a shift in architectural intent. Enterprises who want to hold onto to their customers through positive customer experiences need not ask how they could modify existing systems, but consider what they want the real end goals of their voice support channels to be. Legacy IVRs were designed to route calls and manage volume. Agentic voice AI is designed to resolve customer problems and build brand loyalty.

Fostering brand loyalty through customer support looks less like a menu tree and more like a reasoning system, one that can manage ambiguity, retain context across conversations, complete multi-step tasks without human escalation, and hand off cleanly when human judgment is needed. 

The obituary for the IVR has been written many times, and yet, despite better technology and heightened customer expectations, companies continue to settle for this good enough approach. The reality is, however, that with more choices and better AI understanding from consumers, this approach won’t be good enough for much longer. The writing is on the wall for how to increase customer loyalty through support experiences for the next 10+ years. Those who read it and respond will be the ones who earn it. 

Want the full picture? Parloa's State of Agentic CX report benchmarks how enterprises are delivering customer experiences across voice, chat, and web support globally.

Download the report