The AI situationship: when automation won’t commit to the customer
I recently learned a new word from my kids: situationship.
When I asked my daughter what it meant, she explained it very clearly: if you don’t define it, you’re in a situationship. Nobody knows what it is, what it means, or what to expect.
I laughed. And then I realized how often companies put their customers in the exact same position. In fact, over the past couple of weeks, I had two experiences that made this very real.
The first was with a grocery order. I’ve been shopping with this chain for decades – actual decades! Every week, like clockwork, the process is routine; my orders are consistent.
Last week, one delivery showed up missing more than half the order. While frustrating, that’s not unusual in itself. Grocery delivery has issues. What matters is how it gets resolved.
I did what I have learned to be the most effective: I called. But the company had changed the process, and what I ran into was a fully automated front end with no real path to a person or, as I call it, an “escape hatch.” I had to authenticate, explain the issue to a bot, repeat myself, and then the system decided it understood what I needed. It processed something I didn’t ask for and ended the call.
No confirmation of what was actually missing. No ability to clarify. No way to correct it.
It took three more calls and over 30 minutes to reach a human. By then, my issue wasn’t just the missing groceries but the overall experience. None of the 30 minutes had to happen. The issue could have been resolved in minutes had there been automation that was thoughtfully built to consider all of the problems a customer like me could have been calling about.
Eventually, I was able to speak to someone and get my partial refund. I was still annoyed at the time suck but relieved to have my partial refund. I was ready to move on with my day without completely writing off the grocery store.
Then emails started. The system had logged a request to refund the entire order, something I never asked for. That request was reviewed and rejected by another layer of automation. So now I’m getting messages about a decision tied to a request I didn’t make.
Eventually, the correct refund came through. But it didn’t matter. Thirty years with the same brand, and I’m officially not going back. There are too many other places that do it better.
That’s the situationship.
This is just one example of how so many automation efforts have been failing. The grocery store didn’t think through the right kind of automation, the kind that would have ensured I could get what I actually needed from them. In this case, a relatively simple issue turned into a breakdown across multiple touchpoints. The automation didn’t recognize my intent correctly, didn’t allow for correction, and didn’t give me a clear path to resolution. It was a time consuming fail for a basic query that left me with time I can’t get back and a bunch of annoying emails to sort through.
What they should have done
If I were looking at this from an operational standpoint, the starting point is straightforward.
First, understand why customers are actually reaching out. Not just “refunds” as a category, but the real scenarios. Missing items, damaged items, partial orders. Those need to be explicitly accounted for in the interaction design.
Second, make a clear decision about what automation should handle and where a human needs to step in. Removing access to an agent is a choice. If you make that choice, the automation has to fully support the resolution. In this case, it didn’t.
Third, test it the way a customer would use it. Not just whether the system technically works, but whether it works when someone is frustrated, in a hurry, and trying to fix a real problem.
And finally, monitor what happens after rollout. Issues like this don’t stay hidden. They show up quickly in customer behavior.
None of this is complicated in theory but does require knowledgeable and thorough execution and QA.
Why? Clearly, the cost is real. Saving a few minutes of agent time is not worth losing a customer who has been with you for decades. As a customer myself, I know that what’s most important to me is a system that works the way I expect it to. If I don’t get that, I will leave.
Blue Orbit Consulting has been transforming contact centers for over 15 years. Interested in implementing new technology or AI tools into your customer service operation? Let’s talk.