FedEx’s Use of AI Chatbots Is the Worst Thing a Company Could Do to Its Customers

If you’ve ever had a package delayed by FedEx, you already know the feeling of frustration. You refresh the tracking page for the tenth time, watching as the promised “by noon” delivery window ticks by. Only it’s not coming.

So, you do what people do and try to figure out what’s going on. One of the most amazing technological advances of the last 50 years is that you can watch a package go from Seattle or Los Angeles, travel across the country, and arrive at your home in Miami or New York. The amount of coordination and logistics that go into making that happen is not something I can comprehend. The problem is, sometimes it seems like it’s just theater.

For example, I was recently waiting for a package promised to be delivered by noon, though the tracking information said it never even left FedEx’s hub in Memphis. Even still, it insisted it would arrive at my door on time—despite being 700 miles away.

That doesn’t really make sense, but it’s not nearly as bad as trying to actually contact FedEx’s customer support, which is now an artificial intelligence mess. You’d think that technology would mean faster responses or better answers. What it actually means is that the company has built a wall between itself and its customers—and then put a talking robot in front to tell you to go away.

AI virtual assistants

When you open the chat window on FedEx’s website, you’re greeted by an AI “virtual assistant” that offers to help. It can tell you what you already know: your package hasn’t moved. It can read tracking data, copy-paste policy lines, and assure you that it has the most up-to-date information. What it can’t do is anything remotely useful.

I kept asking the chatbot why my package wasn’t delivered, but it just kept insisting that it was scheduled for delivery that day, even though it was already 10 p.m. Telling me that a package still sitting in an airport in Memphis will be delivered in Michigan is neither up-to-date nor useful.

Of course, if you ask it to talk to a person, it’ll ask you to call. That seems reasonable, but if you do, you’ll hear that “our agents have the same information you can find online.” In other words, the company doesn’t even want you to try. It’s a remarkable statement: not only is the bot incapable of helping you, but FedEx seems proud of the fact that its human employees wouldn’t be able to either.

Humans want to talk to humans

The thing is, I know for sure that the humans can help you. At a minimum they can try to explain what went wrong. In some cases, those humans will go out of their way to try to solve whatever happened to your package.

I mean that sincerely. FedEx has a long tradition of employees going out of their way to help customers get their deliveries, sometimes taking extraordinary measures to deliver a passport or business contract.

And—to be very clear—the problem here is not with the planes and delivery trucks and people who deliver FedEx packages. Sure, my package was delayed, but I fully understand that things happen. I wasn’t mad about it, I just wanted to know what happened so I could plan.

The problem isn’t even with the people you might talk to on the phone if you’re able to figure out the secret pathway through to an actual human. The problem is with the people who make decisions about how to do things like “streamline operations” and “increase efficiency,” by inserting technology in places where humans would rather interact with other humans.

The wrong incentives

Companies like FedEx know exactly what they’re doing. They don’t deploy AI chat systems because customers love them. Companies do it because it’s cheaper than hiring enough people to handle the number of inquiries and complaints they get. They do it because they know most people will give up before ever talking to someone who might actually solve their problem.

And, to be fair to FedEx, it is definitely not the only company that is doing this. I wrote previously about how UPS and Taco Bell are inserting robots where people would prefer to interact with a human.

If you think that you can use AI to save a bunch of money by letting your customers talk to robots instead of humans, I promise you, you’re doing it wrong. Your customers do not want to talk to robots, they want to talk to a person.

I’m sure that there are times when the robots will provide a better answer, but it is not a better experience. And anyone who tries to justify it as being a better experience is thinking about the wrong incentives. It probably seems less expensive, except that it really isn’t when you make enough of your customers mad that they decide they don’t want to be your customers anymore.

The illusion of a better experience

Also, just because your support team ends up dealing with fewer customers doesn’t mean there are fewer problems. It just means that the customers who have those problems gave up before getting them solved. It just means they’re out there getting mad, and that chips away at your brand promise in ways you don’t even realize because you decided you didn’t want to hear from them.

The irony is that FedEx’s business is built entirely on reliability and communication. The company wants you to trust it to deliver something important—something valuable—on time. But when that trust breaks down, the least it can do is acknowledge you as a person. It’s hard to overstate how damaging that is to a brand.

In theory, AI could make customer service better. A well-trained model could predict issues before they happen, proactively communicate delays, and make sure humans step in when empathy or judgment is required. In practice, FedEx has done the opposite. It built a system that’s designed to minimize the number of human interactions precisely when customers need them most.

That might save money in the short term. But in the long term, it teaches customers not to trust you. It teaches them that if something goes wrong, they’re on their own.

I reached out to FedEx, but the company did not immediately respond to my request for comment.