In the previous blog I already shared some pretty bad chatbot examples. I argue that these bad experiences come from bad design choices. Recently I had an even more crazy experience when calling customer service. Let me use that to explain my design reasoning.
I recently needed to travel to Oman by night flight which means I’d land at 10.30 in the morning.
It’s important for me to then be able to go to my room and not have to wait until 15.00 for a check-in.
So I call customer service (because with most hotel websites, you can forget about finding any FAQ’s online) and the following scenario unfolds.
First I get the long, standard (sigh) this can be recorded for quality purposes disclaimer.
Then a very long silence (I literally envision the phone system now needing to switch to the chatbot system and it’s apparently not a smooth transition).
“Hi, my name is Tim” (super clear it’s a computer, not a person) “and I’m here to help you”.
“Please speak slowly when you ask your question.”
Side note: not the best experience already if you need to tell me to speak slowly.
I dictate my question at normal speed to test it.
Another super long silence that almost makes me decide to hang up.
But just in time, I receive an answer.
Granted: not too bad of an answer, saying the earliest check-in is at 2PM.
OK, I now still need to talk to a person, because I can not arrange an early check-in with Tim.
I hang up because I decide to send them an email so I can explain my specific situation and ask for the best solution.
Within 2-3 seconds, I get a call from a fixed number.
Normally I ignore numbers I don’t know, but I assumed this was probably connected to the call I just ended.
I pick up the phone and honestly, unbelievable what happens next.
A human answers by saying: “I was just in the middle of answering your question about the check-in time, why did you hang up?”
Are you kidding me?
Do you really think I’m that ignorant that I don’t know the difference between you and a computer?
I immediately hung up, I was flabbergasted.
In hindsight I would have loved to go into discussion with him, but then realized: he is probably just instructed to say this to me and not the person who designed this experience.
This was a large, global well-known hotel chain by the way.
The only way for me to explain how a company can ever think this is an amazing idea, is when they only design the experience for efficiency.
Ok, and their integrity might not be the highest as well in this case if they are faking the idea that I was talking to a human.
I can already hear the call center and innovation team brainstorming together: how can we reduce the number of calls? Let’s implement an AI chatbot!
Great idea! But what if people hang up the phone while using the chatbot?
Let’s call them back and continue the conversation to give them a great experience!
Let’s not be too honest and tell them we are still learning, we just act as if we were just talking to them!
I know, I know, it’s a bit cynical, but I really think we need to wake organisations up if we want to enhance the experience of customers and employees and also get a clear business impact.
Because yes, business results are still easily enhanced through great experiences.
You just need to design them from a more holistic perspective than efficiency.
Not just from the perspective of reducing “unnecessary” contacts.
Not just from the perspective of the organisation system.
Place yourself in the shoes of your customers (who are not stupid).
Let your employees play the role of the customers when reviewing the experience.
If your employees can still tell you the experience they are testing is not up to par, then keep improving it.
Keep learning, ask feedback from employees first before you start asking feedback from customers.
So when you are thinking about using AI, at least check some of these design principles:
- What problem are we looking to solve (ie. reduce preventable calls)?
- Do we have sufficient data quality to teach the AI and build the model?
- Do we need AI or workflow automation?
- When will this pilot be a business success (not just a sexy toy for PR purposes)?
- What should the experience feel like for our valued customers?
- What should the experience feel like for our valued employees?
- Where can we transparently share that this is an experiment?
Find data to have a baseline measurement and track some data on your design principles to keep learning and tweaking as you experiment with AI.
That’s how you will prevent crazy outcomes as the one this hotel has designed.
This blog is part of a new blog series by dr. Zanna van der Aa
AI and CX: Demystifying the Hype
