Many brands still measure customer experience by whether the issue was resolved. Pasquale DeMaio, Vice President at Amazon Connect, argues that this is no longer enough.
In a recent interview with CX Today, DeMaio said the bigger risk is not always visible failure. It is the customer who gets through the process, accepts the answer, and then quietly decides not to come back.
AWS explored the same theme in DeMaio’s recent blog, “Failure Didn’t Lose Your Customers, Mediocrity Did”. The argument is blunt, and rather useful: resolution may close the transaction, but it does not always earn the next interaction.
AWS’s latest Amazon Connect Customer launches are designed to support that shift from basic resolution to more relevant, context-aware customer experiences. The new capabilities focus on helping organizations build AI-native, agentic interactions that can understand customer intent, use richer context, and guide customers toward more useful outcomes across voice and digital journeys.
For CX leaders, that creates an uncomfortable gap. A dashboard can show lower wait times, higher automation, and more resolved contacts while customers still feel processed rather than understood.
Deflection Can Send Customers Somewhere Else
For years, service teams were told to reduce calls, contain demand, and deflect customers into lower-cost channels. DeMaio said those terms now look awkward when viewed from the customer’s side.
In his view, the language reveals the problem. Customers do not want to be deflected. They want help that recognizes who they are, what has happened, and what they are trying to achieve. DeMaio framed the risk clearly:
“If you're deflecting that customer away, you have to think about who you're deflecting them to. And in that case, is it to an AI search engine? Is it to a competitor? Because both of those folks are going to accept them with open arms.”
The point is not that automation is bad; in fact, AWS is making the opposite argument. AI can improve customer experience when it helps brands understand context, act earlier, and give customers a better route to resolution.
The problem starts when automation is designed mainly to keep customers away from the business.
DeMaio said brands need to recognize customers more deeply, rather than greet everyone with the same generic language. A loyal customer does not need another empty “thanks for being valued” message. They need evidence that the business knows what they bought, what might be happening, and how it can help.
As DeMaio put it, customers can tell when a company is trying to solve their problem and when it is trying to “get rid of them.”
The Quiet Exit Is Harder to Measure
A failed customer interaction usually leaves evidence. The customer complains, escalates, gives a low score, or contacts the company again.
Mediocre service is less helpful and can leave no dramatic signal at all. DeMaio described the customer who receives an answer but remembers how painful the process felt:
“People are willing to crawl across glass to get things done that they need to get done, but they're going to remember that you made them crawl across glass.”
That is the quiet exit problem. Customers may not complain. They may not fill out a survey. They may simply choose another brand next time.
DeMaio argued that traditional feedback tools can miss this middle ground. CSAT and Net Promoter Score can still be useful, particularly when experiences are very good or very bad, but they do not always reveal gradual indifference.
CX leaders need to look across a broader set of customer signals. Are customers still engaging with the brand? Are they returning to the website? Are they trying to solve the same problem again? Are they buying again, calling again, or disappearing from familiar patterns?
The awkward bit is that a “resolved” case may still represent a damaged relationship.
Context Changes the Customer’s Experience
DeMaio used a delayed birthday gift as an example in his blog. A refund or replacement may resolve the transaction, but the customer’s real problem remains if the birthday is tomorrow.
The distinction is simple. The operational problem is the delayed item. The customer problem is the missed moment.
In the interview, DeMaio said circumstances matter. Some purchases can wait. Others cannot. Organizations can infer some of that context from customer data, while in other cases they can ask the customer directly.