If automation adds friction, it’s failing
A few days after this grocery experience (The AI situationship: when automation won’t commit to the customer) that made me break up with the grocer I had been with for decades, I had the opposite situation when I had to cancel two credit cards with two different banks.
The first call took about 10 minutes. I reached a person quickly, which I always advocate for. The agent was professional and thorough. He followed the process step by step, read the disclosures, and did exactly what he was trained to do as part of the card company’s win-back script.
It was longer than it needed to be, and they didn’t win me back. It was fine - what I'd consider an acceptable experience.
The second call was handled entirely through automation. Less than two minutes after I called, the account was closed. I had an email confirmation before I hung up. And here’s the important part: I trusted the outcome. Here's the funny part: I was so happy with this experience that I’d come back to this bank if I were ready to open another credit card.
Two experiences, one outcome
The difference between these two experiences was the alignment between the interaction design and the customer’s goal. In one case, the company followed a traditional model in which the call routed to an agent, there was an attempt to retain the customer, and required steps were followed. It works, but it’s not always necessary.
In the other, the company made a different decision. They recognized that some interactions don’t require persuasion, escalation, or human intervention and that the ROI of having an agent do this work isn’t there. They designed the experience to resolve the task cleanly and quickly.
I’ve also seen the opposite play out recently (link to previous blog), where automation is put in place without fully supporting the customer’s need. The result is confusion, rework, and ultimately lost trust. One approach creates friction while the other creates confidence and loyalty.
Where automation strategies start to break down
Organizations are generally not spending enough time understanding why customers are reaching out in terms of context and the expectation behind the call. Removing the human option is a design decision. If you make it, the automation has to fully support the customer’s need. If it doesn’t, you create frustration very quickly.
Testing is often treated as a technical exercise rather than a user experience exercise, so the system may function as designed but still fail in practice. There is not always enough visibility into what is actually happening once these systems go live, which means when something breaks, it shows up first in the customer experience.
What struck me about these experiences is how small the gap can be between working and not working. One company spent more time and more cost on a human interaction that didn’t change the outcome, while the other reduced time and cost but still delivered a clear, complete experience.
Customers don’t need every interaction to be human, but they do need every interaction to make sense. Otherwise, you’re not creating an efficient experience and customers don’t tend to put up with those experiences for long, let alone come back for more.
Blue Orbit Consulting works with organizations to make sure new technology actually works in practice. That means understanding customer intent, designing the right interaction model, and making sure automation, workflows, and support structures align before and after launch.
If your current experience feels unclear or inconsistent, it’s usually not a technology problem alone. It’s a design and execution problem. And that’s fixable.