Key findings from Parloa’s first-of-its-kind state of agentic CX study

An agent-led, human-augmented benchmark of enterprise CX across web, chat, and voice.

Customer experience isn’t breaking down in the margins. It’s failing at the front door.

To understand the scale of the problem, Parloa deployed bespoke AI discovery agents across more than 10,000 enterprise websites, spanning the Forbes Global 2000 and beyond. These agents then conducted nearly 4,000 “secret shopper” interactions across 800+ companies in 27 industries, predominantly focused in banking, insurance, travel, retail, and health care.

What these agents uncovered goes beyond a B2C tech adoption gap.

Across digital entry points, chat interfaces, and voice systems, enterprise customer experience (CX) is simply not keeping pace with how customers expect to interact today. And more importantly, it’s not built for where interaction is going next.

A system under strain

The research evaluated three core layers of the customer journey:

• Access: how easily customers can find support

• Interaction: how effectively chat systems resolve issues

• Execution: how voice systems handle real-world requests

At each layer, the same pattern emerged: friction, fragmentation, and failure to resolve.

More than 43% of enterprise websites offered no visible phone number or chat option.

Customers arrive with intent and find no clear path forward.

Even when support exists, it rarely works:

• Just 8.9% of chatbot interactions resulted in resolution.

• Only 10.1% of escalation attempts reached a human.

• A mere 7.5% of chatbots showed demonstrable conversational AI capability.

And beneath it all, only 1% of enterprises are structurally prepared for agent-to-agent interactions, the model already beginning to define the next era of CX.

The digital front door has yet to be established

The breakdown starts before a conversation even begins.

Parloa’s agents scored each company on how easily support could be found. The results point to a consistent pattern of inaccessibility:

• More than half of websites required scrolling and clicks to locate support.

• A third of phone numbers were buried in footers.

• 37.8% presented multiple phone numbers with no clear guidance or alignment.

Chat is even more constrained. Fewer than 30% of companies offer it at all. And when they do, it’s often treated as a design element, not a service layer, confined to homepages instead of embedded across actual support journeys.

The result: customers aren’t presented with options for channel of choice. They’re struggling to even find one.

Modern interfaces, legacy systems

On the surface, many experiences look updated: chat widgets, upleveled intelligent voice agents (IVAs), AI-labeled interfaces. Underneath, the architecture hasn’t changed.

The study unveiled many enterprises have layered automation onto systems originally designed for touch-tone inputs and rigid decision trees. The result is predictable:

• Chatbots without memory, resulting in customer frustration

• Voice systems that sound natural but behave mechanically

• Handoffs that erase context and force repetition

This isn’t a failure of AI capability. It’s a failure of system design.

Automation was originally introduced to deflect cost, not resolve complexity and certainly not to develop trust and deepen relationships. And that tradeoff is now visible to every customer who gets stuck with legacy platforms.

Expectations have already moved on

They interact with AI that remembers context, adapts in real time, and completes tasks without friction. Then they call a service line and are asked to “press 1 for English.”

That gap is no longer tolerable, and it’s widening.

The collective cost compounds quickly

Each layer of failure builds on the last:

• If support can’t be found, customers leave. LTV lost.

• If it’s found but ineffective, frustration escalates. Brand equity diminishes.

• If resolution requires repetition, trust erodes. Loyalty evaporates.

In regulated industries, the stakes are higher. Failed interactions don’t just lead to churn, they lead to complaints, disputes, and regulatory exposure.

And a new factor is emerging: personal agent intermediation.

As agent-to-agent interactions become viable, customers won’t navigate broken systems because they have to. Their AI will route around them. Enterprises that can’t interface with these systems won’t compete within them.

What needs to change

This is an infrastructure problem that can’t be swept under a rug with surface-level optimization.

The companies that move forward will stop layering niche improvements onto legacy systems and start rebuilding around a different set of principles. Start with these five:

• Make support immediately accessible – no search, no scroll.

• Replace menu trees with intent-based routing, regardless of channel.

• Preserve context across every interaction and handoff. Memory = personalization = revenue from support.

• Shift AI from triage to true resolution. Purely diagnostic is out, end-to-end is in.

• Design agentic systems that can interact with other agents, not just humans. It’s only a matter of time.

These are not incremental upgrades. They redefine how customer experience operates.

A turning point for enterprise CX

For years, operationally-focused automation metrics looked strong while customer outcomes lagged behind. That gap is now visible, measurable, and worse for the enterprise, it’s unforgivable.

The question is no longer whether enterprises need to evolve their CX systems, it’s how fast they need to to stay relevant.

The Parloa State of Agentic CX in 2026 report lays out where the gaps are, why they exist, and what it will take to close them.

Read the full report to see where your organization stands, and what comes next.