The pace of AI adoption has become a board-level conversation in customer experience, with leaders under pressure to answer how it will reduce costs and automate customer interactions. But expectations around how AI will reshape the contact center can often remain some distance from operational reality.
The issue for contact center teams is whether the organization has identified the right problem, involved the people affected and built the operational foundations to use AI safely.
Stephen Butler-Emmett, CX and AI Specialist at QContact, noted in an interview with CX Today that although the technology is often presented as a rapid answer to several business challenges at once, it can create problems when customer-facing automation is deployed before teams understand the context and complexity of their interactions.
“The reality is that even though the technology has come a long way, a poorly executed customer-facing AI solution can cause a lot of harm to a business.”
The expectation is often that AI will reduce costs, replace people and handle all of a business’s customer interactions. “It often gets framed as a silver bullet that’s going to transcend our business,” Butler-Emmett said.
But a poorly designed automated experience can expose that quickly.
The Risks of Treating AI as a Shortcut
Placing AI into customer-facing workflows can introduce risks when systems have to interpret complex, unpredictable interactions. Customer service interactions are rarely neat, structured data points, as individual customers bring different needs, emotional states and reasons for contacting the business.
As Butler-Emmett pointed out: “People often think it’s very structured data and they can throw AI at it. Actually, customer interactions are highly unstructured. What ends up happening is we rush into these things.”
Such rushed implementations, in turn, have a direct impact on customer satisfaction.
“Customers can become infuriated. You can put them in one of those death spiral loops where they just can’t get answers. The handover to human beings is not seamless and it comes at the wrong points,” Butler-Emmett pointed out.
There is also an internal consequence, as introducing AI systems without a clear explanation of their role and purpose can raise concerns among the people expected to work alongside them.
“As soon as you bring AI into a business, you can create a bit of skepticism. People are fearful for their jobs; they often ask, ‘What does this mean for me?’”
Butler-Emmett uses a simple analogy:
“It’s essentially putting a bouncer in front of your customers and your business. If your bouncer doesn’t know how to read the guestlist well, it can be quite detrimental.”
Customer experience leaders need to decide carefully whether using a tool to automate an interaction is appropriate, how customers will move to a human when needed and what the experience will mean for the brand.
Why Pilots Lose Momentum
Many contact center AI initiatives begin as pilots or proofs of concept. When pilots fail to meet expectations, the impact extends beyond a single project.
“The morale in the business dissipates quickly, at all levels,” Butler-Emmett said. “Managerial levels are not getting the results, things are not working, and then it also harms the people working on the floor. They’re demoralized, and the board’s demoralized with the outcome.”
And beyond that immediate impact, the longer-term risk is a loss of confidence in future technology programs “for years to come,” Butler-Emmett said, “whether it be approval for funding and budgets, or just the appetite to do it again.”
Rather than treating AI as a broad customer-facing transformation, QContact suggests a more disciplined approach that focuses first on areas where the technology can support staff and improve their understanding of customer demand.
Start With Internal Support
Effective early use cases tend to sit in the back office, Butler-Emmett explained, including analyzing contact volumes, identifying recurring themes, supporting quality assurance and helping agents find information or shape responses.
