AI's Real Promise: Making Customer Service Feel Easy

New Avaya research points to a powerful opportunity: less to repeat, less to chase, and a clearer path to getting help.

A few days ago, I had a few minutes between meetings and thought I could squeeze in a simple task: scheduling an appointment.

A chatbot took me through a lengthy list of questions before transferring me to a person. I hoped the person would already have the information I’d just provided. Instead, I had to repeat everything.

I got the appointment. I also arrived a few minutes late to my next meeting.

If all we measured were whether the appointment was booked, we would call that a success. The extra time and effort tell a fuller story.

New Avaya research points to a powerful opportunity: less to repeat, less to chase, and a clearer path to getting help.

Imagine two customers contacting a company about the same billing mistake.

Both get a quick response. Both encounter a polite AI assistant. Both need help from a person.

One customer hears, "Can you explain the problem?"

The other hears, "I can see what you've already tried. Let's pick up from there."

On a dashboard tracking response times and transfers, these two journeys could look remarkably similar. To the customers, they could feel like different companies.

That small difference offers a useful way into Avaya's new report, Customer Experience on Their Terms. In our September 2026 consumer research, only 1 out of 2 US consumers said their most recent customer-service issue was resolved during the first interaction. Most others finally got a resolution, but needed more than one interaction.

For leaders investing in AI, the opportunity extends through all those interactions. How much easier could getting help become if each step built on the one before it?

Where Does Customer Effort Hide?

Often, it hides in the gaps between otherwise useful interactions: repeating an explanation, finding the right person, or checking whether a promised action happened.

In the Avaya survey, 49% cited transfers between people or departments as a source of difficulty, while 47% cited having to repeat information. Respondents could select up to three sources of friction.

The revealing detail is what those frustrations share. Both concern what happens as service moves from one step to the next.

An assistant can identify a billing issue correctly. An agent can know how to resolve it. Connecting those capabilities, with the relevant information intact, is what makes the journey easier.

Continuity is a practical form of convenience. It lets a customer keep moving without rebuilding the conversation.

Qualtrics' June 2026 research, based on 7,001 consumers across seven countries, adds a telling detail: AI interactions received stronger ratings for friendliness than for understanding the customer's issue. A pleasant conversation can still leave someone needing help.

The implication for AI design is straightforward. Judge the experience by whether the customer's need was addressed and how much effort it took to get there.

What Do Customers Want When AI Cannot Solve the Problem?

They want access to a person who can continue the work already started. This leads to a useful insight: an AI assistant can create value by knowing when, and how, to bring someone else into the conversation.

In Avaya's research, 82% wanted a human option after several unsuccessful attempts with an AI assistant. Specifically, 64% of all respondents wanted an automatic connection to a person with the conversation and information carried forward. Another 18% wanted the AI to ask whether they would like a person.

The handoff becomes part of the AI experience itself. An assistant that recognizes its limit and prepares the next person to help can save the customer another round of explanation. Knowing how to continue is a capability worth designing for.

The expectation extends across channels. 91% wanted previous conversation context carried forward in some form when changing channels or devices. Of all respondents, 67% preferred this to happen automatically once their identity was verified. Others wanted to be asked or to request it themselves.

For service leaders, that suggests a simple test: can the next person see why the customer contacted you, what information they already supplied, and what has been tried? Relevant context should travel with appropriate identity checks, permissions, and access controls.

The human role also deserves a rethink. In the survey's closing question, 72% selected easy access to a knowledgeable person when asked what would most improve service. They could choose up to two changes. The word knowledgeable matters. Access is only useful if the person is equipped to help.

Gartner's April 2026 survey release reports that 85% of service and support leaders were expanding human agent responsibilities. As those responsibilities grow, preparation, knowledge, and access to the right tools become more consequential.

Do Customers Actually Want AI to Act For Them?

Yes, many do, but the task and the permission model matter.

Avaya asked respondents to imagine that a company's AI assistant detected a duplicate charge. 91% were open to AI playing some role in addressing it. That included notifying them so they could fix it themselves. The largest group, 37%, wanted AI to prepare the correction and ask for approval.

Read alongside the preference for human help after repeated AI failure, this points to something practical. Customers can welcome AI assistance and value a clear route to a person. The appropriate role depends on the situation.

It also depends on permission. Interest in AI assistance does not imply consent to every action it could take.

Pew Research Center's September 2025 findings add perspective: 73% of U.S. adults were willing to let AI assist them at least a little, while 61% wanted more control over its use in their lives. These are broader attitudes toward AI, separate from Avaya's customer-service scenarios.

Control is part of usability. People need to understand what an assistant can do, when it will ask, and how to get help if something goes wrong.

For a duplicate charge, a well-designed experience might identify the transaction, explain the proposed correction, request any required approval, and confirm the result. The customer makes the decision that matters, while the system coordinates the steps around it.

How Should Leaders Measure AI Customer Service?

Start with completed outcomes, then examine the work required to reach them.

Alongside response speed and contact volume, track:

• Repeat contacts for the same need. Did the customer have to return after the interaction was marked complete?
• Time to confirmed resolution. How long did the whole process take, including transfers and follow-up?
• Handoff quality. Did the next person or assistant receive enough relevant information to continue?
• Customer effort. What did the customer still have to repeat, chase, or coordinate?

These are suggested measures, not additional findings from the survey. Together, they help reveal where automation makes the whole experience easier.

Start with one common journey, such as a billing correction. Follow it across every team and system involved. Look for the moments when progress pauses because information, authority, or responsibility needs to move.

Those moments can reveal opportunities that an individual conversation score would miss.

How Can AI Make the Whole Service Journey Easier?

The business needs to connect information, decisions, and follow-through across the whole service journey.

McKinsey's August 2026 State of AI survey found that nearly three-quarters of its AI high performers reported fundamentally redesigning workflows, compared with one-quarter of other respondents. The association does not establish causation, but it reinforces the case for examining how work gets done alongside the technology used to do it.

In customer service, that means connecting the conversation to account information, the next permitted action, the people who can help, and confirmation that the work was completed. This coordination is what we mean by customer experience orchestration.

That is the role Avaya Infinity is designed to play. Infinity brings customer context, AI, enterprise systems, workflows, and human expertise together so the next step can build on what is already known. Its role is to help coordinate the resources behind the experience, including transitions between automation and people.

Return to our two imagined customers. The second conversation feels easier because the person helping already has a place to begin. Extend that continuity through the whole journey: preserve the explanation, bring the right information to the agent or AI, apply the appropriate approval rules, and follow the correction through to confirmation.

There is considerable coordination behind an experience that feels simple. AI's opportunity is to help handle more of that coordination so customers can get on with their day.

The test for Infinity, and for any customer experience investment, should remain the same: how much easier did we make it for the customer to get the right help?

Read the full Avaya report, Customer Experience on Their Terms, for the findings on resolution, continuity, AI permissions, and human expertise.

About the Research

Avaya's Customer Experience on Their Terms is based on an online consumer panel survey fielded in September 2026 among U.S. adults aged 18 and older.

External Research Referenced

• Qualtrics, June 18, 2026. Research on agent performance across AI and human channels. Based on the 2026 Agent Effectiveness Benchmark Study, fielded December 2025 through January 2026.
• Gartner, April 28, 2026. Survey findings on expanding human agent responsibilities. Survey of 321 service and support leaders worldwide, fielded September through October 2025.
• Pew Research Center, September 17, 2025. AI in Americans' lives: Awareness, experiences and attitudes. U.S. adult survey fielded June 9 through 15, 2025.
• McKinsey, August 25, 2026. The state of AI in 2026: On the road to ROI. Global survey of 1,719 respondents, fielded May 4 through June 8, 2026.