If this was fwd’d to you, subscribe here.

MOTW (Meme of the Week)😀

☺️ Hey, Hey, Heyyy! ☺️

AI customer service is getting dragged right now.

And honestly, some of it is deserved.

CNBC recently covered the growing frustration with AI customer service, including research showing that nearly 1 in 5 consumers who used AI for customer service said they saw no benefit from it.

Then there was the Forbes article literally titled “I Called Customer Service. All I Got Was AI Slop. So, I Canceled.”

So brands are being pushed to use more AI at the exact same time customers are getting increasingly annoyed by it.

But I don't think customers actually hate AI.

They hate bad AI.

Here are three things brands need to get right:

1. AI has to actually be set up correctly

This seems obvious, but apparently it isn't lol.

If your AI doesn't have the right information, doesn't understand your policies, or keeps sending customers in circles, it's not saving anyone time.

Imagine being the customer in that situation. At some point, you don't care how advanced the technology behind the chatbot is. You just want your $10 back 😭

Before you worry about how many tickets your AI is deflecting, go through the experience yourself.

Ask it weird questions. Give it incomplete information. Ask for a refund. Try to reach a human. See what happens when it doesn't know something.

Try to break your own AI before your customers do.

2. Your AI still needs a human managing it

AI customer service isn't something you set up once and never look at again.

Someone needs to be checking what it's saying, where customers are getting stuck, what questions it can't answer, and when people are asking for a human.

That is a HUGE disconnect.

Your dashboard might tell you the AI is working perfectly while your customers are having a completely different experience.

You still need a human paying attention.

3. Know when AI should NOT be answering

This is probably the biggest one.

AI is amazing for order status, basic FAQs, return policies, product questions, gathering information, summarizing conversations, and all of those repetitive tickets your CX team answers 900 times.

But not every customer interaction needs to be automated.

The Forbes story is a perfect example. The author had a relatively simple account issue, couldn't get useful help through the company's automated support, received what he described as an AI-generated response afterward… and eventually canceled.

That's where brands need to be careful.

If someone is angry, has a complicated billing problem, is a high-value customer, or has already tried your AI twice without getting an answer…

PLEASE just let them talk to a person 😭

The goal of AI in CX shouldn't be to eliminate humans or deflect as many tickets as physically possible.

It should be to make getting help faster and easier.

Because saving a few dollars on a support interaction isn't really a win if the experience costs you the customer.

Especially going into BFCM, don't just turn on more AI because ticket volume is coming.

Make sure it's actually good first.

💬 Question of the Week

What's the worst experience you've had with an AI customer service bot?

I KNOW everyone has one.

😌 Personal Update

I’m doing a bit of traveling and recording some Let’s Laugh About It podcast episodes along the way! 🎙️

If you work for a brand, are interested in recording, and will be in any of these cities, reply to this email! Or, if you know someone who’d be a great fit, send them my way.

Chicago — Sept 3rd
NYC — Sept 13th–16th (Anyone going to Beanstalk? 👀)
NYC again — Sept 22nd–24th

🎙️ LET’S LAUGH ABOUT IT 🎙️

New Let's Laugh About It episodes are coming 👀

We're talking to founders, operators, and CX leaders about the things that went wrong, what they learned, and the stories that are (hopefully) a lot funnier now.

🚨Inevitable Agency CX Audits 🚨

If you're adding AI to your CX, getting ready for BFCM, or aren't totally sure what's actually happening inside your customer experience, we can take a look.

Our CX Audit digs into your customer journey, support operations, AI and automation, and where things might be going wrong.