My package was marked “delivered” by a system that never delivered it. That’s when I knew I was about to lose an afternoon to AI customer service that couldn’t admit it was wrong.

That kind of thing doesn’t just cost you a customer’s afternoon. It costs you something harder to get back.

That’s a brand tax you’re paying without realizing it.

You’re still responsible for how your brand acts and communicates, even if it’s a bot doing it. So, you’d better get it right.

Recently, I’ve had a few instances in which I needed to talk with a live human regarding issues with invoices received. Each one added to the tax.

First was a well-known shipping carrier. We received a notification that our offices were closed and the urgent item we needed couldn’t be delivered. However, our offices weren’t closed.

Our CEO went over to the local shipping carrier site where the packages are held and found out that the delivery driver hadn’t shown up to work, so all deliveries were just marked as “business closed.” Wow.

After getting a manager involved, our package was unearthed and picked up at 4:18 pm. Then we received our invoice. Since we picked up the package, it was marked as “delivered” in the system, and we were charged the full amount for the delivery.

But wait a minute. The carrier didn’t deliver the package; we had to pick it up. So, I went to dispute the charge through the billing portal. Except there wasn’t an option for the reason we were disputing the charge. There’s a “helpful” AI assistant chat in the portal, which didn’t provide any useful information for our particular issue.

Next, I made a phone call, where, guess what? I got an AI voice system. I went through the voice prompts to dispute a billing, and because the system was showing “delivered” in the time promised, the dispute was denied. After asking for a representative, I eventually talked to a real person. But the system wouldn’t listen to the real person either. The issue had to be escalated, and we’re awaiting the results. What a terrible customer experience.

Another telecommunications vendor was showing a different balance due in their billing portal than the last invoice showed. I couldn’t find any changes that accounted for the difference; the AI chat kept going around in circles, and when it offered a connection to a live representative, it just started the AI responses all over again and never connected to a real person. I’m just going to have to wait until the next invoice is generated to find out what happened.

These two recent negative experiences have soured me on both brands. And I’m not the only one. A recent survey shows that 73% of consumers indicate a negative customer service experience greatly impacts their overall perception of a company. Of their brand.

I have also had some favorable interactions with AI customer service chats, but they were mostly for returning items online. The interactions were easy, quick, and didn’t impact my views of the company negatively. However, if an issue arose during the return and I didn’t get to a live person, that positive interaction could go south quickly. AI earns its keep when it quietly gets out of the way. It costs you the moment it stands between a customer and the person they actually need.

A 2025 study from Kinsta found that 81% of consumers believe AI is used primarily to save money, not to improve service, and 41% say that customer service has gotten worse due to AI. Is that the message you want your customers to receive?

None of this means AI has no place in customer service. My return experiences prove it can work well for simple, low-stakes interactions. But the irony is that AI is supposed to make customer service more efficient. Instead, in my experience, the moment something didn’t fit the script, it added friction, created disputes that shouldn’t exist, and burned the goodwill both companies had built over years of otherwise decent service. Call it brand stewardship running in reverse.

If your AI system’s success metric is “fewer calls to a human,” you’re optimizing for the wrong thing. Optimize for “customers who still trust us after something goes wrong.” That’s the number stewardship is actually built around, and the one that protects your brand.

If you’re a business leader deciding where to deploy AI in your customer experience, ask yourself the question those consumers are already answering: are we using this to serve people better, or just to spend less on serving them? Your customers can tell the difference. And 73% of them will remember which one you chose.

Brand stewardship isn’t a slogan we put on a slide. It’s a discipline: treating every automated interaction as a decision about how much trust you’re willing to spend to save a little money.

As a famous song from the 80s says, “We all need the human touch.” It was true then. It’s true now. Good brand stewardship never bets against it.