These problems existed before AI, and beyond Korea.
Gartner has studied avoidable contacts caused by product issues. In Medallia’s 2026 global CX study, 30–40% of departments took no action on customer insights. While 66% of CX practitioners saw improvement, only 17% of consumers agreed. Medallia research report
1. AI made the analysis easier, not the decisions.
Analyzing thousands of unstructured conversations used to be difficult. LLMs make patterns easier to find, especially when transcripts connect to product journeys and service data. But it won’t decide which issues matter most or who needs to work on them.
2. Making the business case for a customer problem.
That is what stood out to me in United Airlines’ work with Cresta. Its Insights to Action team estimates the value of a fix, works with the teams responsible, and verifies the results.
Customers were calling because the flight-change self-service flow asked them to pay for an upgrade before knowing whether their preferred change was possible. United changed the order: check first, pay afterward.
Cresta reports a 50% drop in live contacts for that issue and millions of dollars in annual savings. What stood out to me was how the team put a business value on the problem and used it to get the right departments involved. Cresta case study
3. What if we stopped treating incoming calls as inevitable?
Customers call after something goes wrong in a product or service. The company pays to handle the contact. The customer pays in time spent searching, retrying, waiting, and explaining again.
Each department sees the journey from its own side. The feature works, the policy was followed, and the channel met its target. Yet the customer may still be stuck between them, with no problem visible on any one team’s dashboard.
Shouldn’t we also ask whether improvements across the business reduce repeat contacts from the same causes? Not every call can disappear, but repeated causes shouldn’t be ignored.
I’ve worked through these issues one by one. Each time we had to trace the cause, talk to other teams, agree on priorities, and check the result. Then the next issue arrived.
This work shouldn’t depend on one department repeatedly pulling everyone together. We need a way to judge the business impact, agree on what each team can change, and check whether the customer’s original problem was fixed.
This isn’t a new kind of problem solving. It’s the familiar work of dealing with causes, not just symptoms. AI can help us gather the evidence, but organizations still have to act on it.
A contact center handles problems after they happen. It can also be the first place a company sees what needs to change, not merely the last stop in the journey.
If you’ve been thinking about this too, I’d like to hear your perspective.
#ConversationIntelligence #InsightsToAction #CustomerExperience