Insights

When automation breaks trust: 5 CX lessons from a simple receipt request

I’d just booked a flight. All I needed was a receipt so I could expense it back. Straightforward, you’d think.

Finding no joy in the portal, I opened the airline’s live chat and typed my request. The first reply set the tone: “Ok, thank you. My digital brain is still learning, and I am unable to answer this query at this time. You might be able to find what you are looking for in our updated FAQs.”

From there, the spiral began. I clicked through menus, said “No, I still need help,” tried “Speak to an agent”. Each time, the bot circled me back, sometimes to FAQs, sometimes to disability assistance options I hadn’t asked for.

By this point I wasn’t just frustrated, I was wary. If I quit, would I even get back in the queue? That fear kept me stuck, repeating “Speak to agent” like a password, waiting for the system to finally let me through. It took about 15 minutes before I finally reached a human.

The first agent (yes… the first) apologized, read back the thread, and then told me my booking reference didn’t match her system. Another transfer. Another wait.

The next agent was kind and empathetic, but boxed in by policy. And then came the absurdity: “May I know who Chris Mounce is, since you started the chat with the same name for security reasons?”

I’d already given my full name, my booking reference, and explained the request. But because the booking said Christopher and I’d signed in as Chris, the process stalled. On paper that’s ‘data protection’. In practice, it was compliance overtaking common sense.

After half an hour, I finally had an outcome: “You can get an invoice post-travel only. Within 28 working days”.

To claim it, I was directed to a webform that looked like it hadn’t been touched in years – outdated design, no reassurance, no trust. After all that, I was meant to believe this creaky page would deliver the single document I needed.

This wasn’t a niche request. A receipt is as basic as it gets. And if this is the journey for something so simple, what confidence can a customer have in the rest of the service?

5 CX lessons for leaders

This experience is a case study in how a simple, repeatable process can collapse when automation, policy, and design aren’t aligned. There are broader lessons here for any CX leader:

1. Automation without empathy is a trap

A bot that can’t handle routine queries and keeps recycling menus doesn’t save costs, it multiplies frustration. Customers don’t want to feel like they’re outsmarting your system just to get help.

2. Business-first policy creates customer churn

Receipts only after travel might make sense for internal compliance, but to the customer it can signal indifference. Plus, in the context of a company that will also tailor to a business audience, this is a huge oversight – how many customers is this process affecting?

3. Fragile journeys make anxious customers

I stayed in the loop not because it was working, but because I feared being ejected. When customers feel they can’t safely exit and return, you’ve created a brittle experience that erodes trust – and it’s not an experience they’ll wish to repeat.

4. Agents pay the price of bad design

Both humans I spoke to were courteous and empathetic, yet powerless. They carried the weight of flawed systems, apologizing for processes they couldn’t change. Even the ‘Chris vs Christopher’ detour showed how rigid compliance rules undermine agent credibility, as well as customer patience.

5. Trust is designed, not assumed

Handing customers a five-year-old form at the end of a 30-minute ordeal doesn’t inspire confidence. Every touchpoint – especially the final outcome – needs to look and feel trustworthy. Otherwise, you’re once again left providing a terrible customer experience.

Seeing the bigger picture

One bad experience is an anecdote. But how many customers are stuck in the same loop every day? How much wasted time is there on both sides of the conversation?

That’s where Conversation Intelligence and Auto-QA can have the biggest impact. Instead of waiting for complaints or survey comments, you can surface patterns like:

Auto-QA can flag every single one of those interactions without needing humans to manually review transcripts (a great use of automation). You can then get a clear picture of scale: is this a one-off annoyance, or a systemic issue costing thousands of hours of customer and agent time?

With that insight, you can quantify the cost of broken automation versus the value of fixing it. You can see where agents are apologizing for processes they can’t control. And you can identify fragile points in journeys where customers are stuck repeating “speak to agent” for minutes on end.

Closing the loop

If your customer has to fight through 30 minutes of bots, transfers, and outdated forms just to get a receipt, you don’t have a one-off issue, you have a systemic risk.

The details may differ, but the pattern is the same: automation that frustrates, policies that alienate, and agents left apologizing for both. The danger is scale. For every customer who complains, dozens more grit their teeth and quietly decide to fly with another airline in the future.

This is why Conversation Intelligence and Auto-QA matter. They let you hear these moments in bulk, not in hindsight. They show you where loops form, where trust erodes, and where empathy is undermined by process. With that clarity, you can fix the root causes of frustration – and in this case, look to offer a self-service tool that would likely save hours of back-and-forth for agents.

If customers dread asking for something as basic as a receipt, what happens when the stakes are higher?

GUIDE

A complete guide to QA automation

Want to drive efficiency, performance and CX? We’ve got you covered. Make sure you avoid the key pitfalls and get automation right with our free guide.

Download your copy