Reflections following the 2026 Qualtrics X4 conference

A few weeks ago, Rob lost his mobile phone and wiped it remotely. Upon finding it, however, he then spent a morning trying to restore data to his phone and get it working again with his carrier. While restoring the data was no problem, reestablishing the carrier connection turned out to be tricky. And the chat service experience he tried became a circular loop. He couldn't get help without signing in. But he couldn’t receive text messages for two-factor authentication because he had wiped the phone. And he couldn't fix the phone without help.

Rob explained this dilemma to the chatbot, which offered the same three options he had already tried. He asked for a human chat agent. But before he could be connected to a human, the system wanted him to sign in. Eventually, he was able to connect with a human agent. But the agent could not activate his service. After two hours of slow chat interaction, he finally gave up and headed to a store.

There, a technician fixed it in under five minutes.

What’s worth examining is the internal logic of each failure. The bot did exactly what it was designed to do: Route common issues through self-service. The agent did exactly what she was trained to do: Follow the troubleshooting script. Each individual piece of the system was optimized. But they were optimized for cost and efficiency, not customer value.

Jason Maynard, the new CEO of Qualtrics, opened the company’s 2026 X4 conference by describing what most customer experience practitioners know but rarely say out loud: Companies have invested heavily in capturing customer feedback and generating insights, yet most of that signal never informs the decisions that shape what customers actually experience. Maynard framed it as the central problem Qualtrics intends to solve.

To us, the cause runs deeper than a feedback gap, and artificial intelligence (AI), deployed the way most companies currently do, is about to make the underlying condition worse.

The problem with optimizing locally

Here's the pattern we’ve seen for decades. A management team makes a series of reasonable decisions, such as reducing service costs, broadening acquisition targets, streamlining the renewal process, or automating the first tier of contacts. Each decision is defensible in isolation. Yet over time, the cumulative effect causes the company's customer base to quietly deteriorate. Retention erodes. New customers behave differently than the ones they replaced, with shorter tenures, lower spend, and higher cost to serve. Growth becomes harder to sustain while cost-cutting becomes the one reliable lever to pull. No one can point to the decision that caused the doom loop because there were hundreds of small, incremental decisions. None of the decisions truly weighed the likely impact on the future value of the customer relationship.

With Rob’s telecom carrier, the contact center and chat teams each focused on their primary goals of making less expensive automated self-service the easy path for customers while minimizing access to the more expensive human channels. The customer value at stake when reestablishing service is big enough to merit more attention. The telco should have made it easier for Rob to reach a human and should have helped that human agent figure out quickly that she couldn’t help Rob over chat.

So why does this pattern persist? Typically, the management system does not make the trade-off visible. Immediate effects of each decision show up in this quarter's P&L, often in ways that get rewarded. Longer-term effects, such as the customer who stopped calling back because it wasn't worth the effort and the renewal that didn't happen a year later, emerge slowly over time and are nearly impossible to attribute. So the organization learns the wrong lesson. It sees the short-term gain, blind to the long-term cost, and repeats the behavior. The decisions erode the future value of their customer base.

How AI makes it worse

AI doesn't fix this. Worse, it enforces the existing logic more consistently and at greater speed.

Walking the X4 floor this week, we could see technology that will soon make Rob’s telco story obsolete in specifics while preserving it in structure: faster detection, real-time routing, and automated resolution of the cases that fit the script. While impressive, in most companies it is being layered onto the same objectives, metrics, and incentives that produced the frustrating frictions in the first place. A management system oriented toward cost per contact will use AI to drive cost per contact lower and at greater scale. The short-term gains will be real, but so will the effect on customers whose problems don't fit the script, with consequences for the long-term value of those relationships. Here, again, the gains will show up immediately while the damage will surface later, across millions of interactions, with no clear cause.

This is precisely what online travel company Booking Holdings aims to avoid. For its Booking.com website, the company has been reworking its service infrastructure, not to minimize service cost but rather to build service into a competitive moat. Their executives think about customer value as a function of revenue, wallet share, and service expansion, with service cost as an investment to drive that growth rather than a line item to reduce. Their AI deployment follows that logic directly, as they direct AI tools to reduce friction, improve sensitivity to context, and handle customer issues with more precision and empathy. Early results include improved resolution times and higher customer satisfaction scores, with reduced live agent contact rates across brands. Customer service costs per reservation fell roughly 10% this past year, even as booking volumes grew at a double-digit rate. (A longer conversation with Booking’s service leaders will be featured on the Customer Confidential podcast in a few months that goes deeper on the strategy behind those numbers.)

What makes Booking distinctive is not the technology; it’s the philosophy to deploy technology around growing customer relationship value. Every major company at X4 this week has access to roughly the same tools. Booking’s edge is an analytical framework that connects service decisions to the value at stake. The management team evaluates which investments strengthen the long-term economics of their customer relationships and which ones simply reduce this quarter's expense line. That infrastructure makes it possible to point AI at the right objective.

Most companies lack such a perspective. They may have sophisticated P&L management and have excellent customer feedback systems. What they need to do is connect those two things to see, with regularity and precision, what their operating decisions do to the future value of their customer base.

That's the gap Qualtrics’ Maynard gestured at on Monday morning: the path from signal to value-creating decision. Until that gap closes, AI deployment will follow the path of least resistance toward cost, speed, and volume. It will deliver on those objectives better than any previous technology. And for companies managing customer base value, that's an enormous opportunity. For companies lacking that focus, it serves to optimize the wrong thing faster than they ever could before.

Rob’s telco’s chatbot didn’t fail because the technology was bad. It failed because someone decided that routing common issues through self-service was the right objective, measured success by deflection rate, and built a system that delivered exactly that. Better AI will make the deflection rate higher. It won't fix the 11-minute silence between chat agent responses or eliminate the trip to the store; those are symptoms of what the system was designed to optimize, not bugs in the implementation.

The companies that pull ahead will be the ones that measure how their decisions impact the value of their customer relationships over time. They will view success in terms of lower churn, greater share of wallet, better price realization, and higher cross-selling rates. And they will use AI to strengthen those relationships rather than just to reduce the cost of managing them. The others will also improve their numbers, just not the ones that matter most.