Customer experience management in the age of agentic AI

“I am excited to introduce our next step in this journey,the Webex AI agent. This is in pilot right now, it will be available to our customers in an early calendar quarter in 2025. This AI agent brings together conversational intelligence and generative AI to deliver natural conversations with hyper-personalization.”

Those words came from Cisco, the enterprise digital communications and technology company; specifically from Anurag Dhingra, SVP and GM of Cisco Collaboration. The words sounded strangely familiar. In the same month of October, we had heard about similar initiatives from Oracle, which also primarily serves the enterprise, as well as from Zendesk which serves mid-size to enterprise customers and has a narrower offering very much focused on the customer experience.

Something was in the air. On the one hand, the aim of automating the customer experience (including automating the work of service reps) seemed very ambitious. At the same time, if everyone starts doing it, it will soon be table stakes.

But is it a realistic aim? And is customer experience really reducible to the service (or support or success) experience? We spoke to an expert, Isabelle Zdatny, head of thought leadership at the Qualtrics XM Institute.

Defining experience management

First, its useful to clarify the relationship between Qualtrics, an experience management software vendor, and the Institute. The first thing to know is that the Institute is product-agnostic. “Our role is like a think tank inside Qualtrics,” said Zdatny. “We are focused on CX and EX [employee experience] professionals and providing them with the insights, practices and principles they need to be successful in their role — to grow their personal skills as well as help to build an effective and sustainable experience management program. Unlike other internal experts at Qualtrics, we’re less client-focused and more focused on category building — what is experience management and what do people need to know to do it well?”

The XM Institute was formed in 2018 when Qualtrics acquired Temkin Group, a boutique consulting firm founded by Bruce Temkin, who Zdatny calls “the godfather of customer experience.” She had been with Temkin Group since 2013.

Zendesk’s intention of providing AI agents across all channels, working either autonomously or in harness with humans to field customer queries, is an approach to what it calls “customer experience management.” Certainly, Zendesk believes that customer experience “refers to all the interactions between a business and its customers,” but its explanation of its new capabilities always seems to circle back to the call center or to digital customer service channels; the same could be said about the announcements from Oracle and Cisco. Isn’t that perspective narrow?

“It absolutely is,” said Zdatny, “because that’s just reactive fire fighting. Experience management is about more than responding to customer complaints. You have limited resources as an organization. There are probably thousands of problems you could go fix. Experience management is going to help you figure out which ones to focus on, where you should be allocating attention and resources. And it’s not just about fixing what’s broken,” she emphasized. “It’s about how we are delivering the emotionally engaging, innovative experiences that will help us to stand out in a crowded market.”

How does Zdatny think about customer (and employee) experience? She had two definitions, first a “scholarly” one used internally at the Institute; the second, one used in conversation with the C-suite. First: “The discipline of driving actions using an ongoing flow of insights into how customers and employees are thinking, feeling and behaving. It is a systematic business practice, not a set of isolated activities.”

Second, and more simply: “Understanding and optimizing the experiences of customers and employees.”

How should experience management evolve within organizations? “What we see in early stage organizations is a fragmented approach,” said Zdatny. “Product, sales. “What makes for a good program is a centralized group that is able to consolidate and coordinate experiences across the entire organization; you call a contact center or walk into a store, you are having the same type of experience.” For very small companies, she said, centralizing responsibility for experience in one person can work just fine.

Key to optimizing the customer experience, as Qualtrics has long emphasized, is gathering feedback. That isn’t necessarily straightforward. “Early stage CX programs focus really heavily on feedback. Unfortunately, they don’t take a lot of action based on that feedback. They’re collecting a lot of insights but not using those insights to make changes,” explain Zdatny. “An effective CX team has that data and analytics but also other supporting functions like experience design and change management. Feedback is foundational but you can’t have an effective experience program if the insights are just being passed over the wall for other teams to deal with.”

Dig deeper: Zendesk saturates CX with AI and voice

The road to fully the automated experience

Perhaps the simple solution to centrally organizing and continually optimized CX, whether in the broad sense or the narrower sense of support and service, solves for a lot of these challenges. That’s a view enthusiastically advanced by Jeff Wartgow, VP product management, Oracle CX service. But he agrees it’s a matter of advancing along a continuum.

First comes improving service by switching from traditional chatbots to conversational AI. Second, improving the performance of human service reps by providing them with AI assistance (or, indeed, assistants). Third and fourth: improving execution by automating service processes and transforming service with automated execution plans. I asked Wartgow to distinguish between the latter two.

The third stage is: “I know how to fix this, I’m going to automate how we fix it.” The fourth: “What if I don’t know how to fix it, can I automate the planning for how to fix it?” In other words, it’s the difference between using AI to automate a known response to a service issue and using AI to figure out the response to a service issue.”

One challenge businesses will face in pursuing this path is that it will need to have its knowledge base in order for the AI agents to be successfully trained. Two years ago, said Wartgow, Oracle completely rebuilt the knowledge base in Oracle Fusion Cloud. “Say there are 15 service requests and we fixed them all the same way. Shall we just turn that into a knowledge article? You just press a button and genAI will write the article and put it in the knowledge database.”

Wartgow agrees that the knowledge base will also need to be constantly refreshed and says that Oracle has a mechanism to “put fresh water in the fish tank.” Oracle’s knowledge base can also ingest large quantities of legacy knowledge, even hidden in large manuals, and create knowledge articles tuned to specific tasks. “We had to do all this first before we could even start talking about these agents,” he said. Oracle will encourage clients to use the Oracle knowledge base rather than some internal alternative as the main source of truth for Oracle service agents.

Dig deeper: Oracle aims to automate the complete customer service lifecycle

The holistic experience

When asked, Oracle, like Zendesk, will agree that the customer experience is not reducible to the service experience. “I’m the service guy at Oracle,” said Wartgow, “so I talk a lot about service. But 70% of the interactions a customer will have, whether it be B2C or B2B, will be with the service department. But I don’t feel like I am talking to a brand’s service center, I feel like I’m talking to the brand. I should be able to change from a sales, to a service, to a marketing conversation as fluidly as possible.”

Oracle and Zendesk, then, have blueprints for an almost entirely automated future, at least for the service part of customer experience. How will that sit with a world in which customer experience is more holistic than that? “Consumer concern about having a human to connect to is the only concern that went up over the last year and it was over 50% that were concerned,” said Zdatny, referring to XM Institute research (registration required).

Indeed, she points out that there are regulations in Europe that say you have to make it easy to reach a human. “I understand from the company’s perspective it’s more efficient if you’re deflecting calls away from high cost call centers. Right now, consumers are saying pretty clearly that’s not what they want.”

But Zdatny agrees that it’s hard to say what consumers (or B2B clients) will want three years from now, if the AI agents get really good at their jobs. “Over the long term, that is the direction we’re moving. In the short term, I think a lot of companies are out over their skis,” she said.