Automated failure
Picture the scene: it is late on the Sunday night of a busy bank holiday weekend in a popular hostelry on the outskirts of our nation’s capital.
The last round of clientele is preparing to depart and taxis are in high demand. The proprietors are using a popular ride hailing app that co-opts various local taxi companies, but a major sporting event, combined with the bank holiday weekend, has meant that the system is being a bit unhelpful.
Despite the ubiquity of smart phones among the assembled staff and impatient customers, many of whom have the client version of the same app, the lack of mobile signal is hampering things, and the establishment’s Wi-Fi is now taking a hammering too due to the relatively poor broadband service available to this slightly out of the way, but by no means remote, business.
“These combined capabilities leverage the best of machine capability, where patterns are sought and identified in large amounts of data, to escalate them to humans for ultimate determination of the best way to rectify them”
Experienced in such situations, the management keep everyone happy with regular updates and a continued table service, even if the updates are mostly that they are still trying.
Frustrated, but in general good spirits due to a most enjoyable evening passed with fine company in said establishment, this hack did what everyone else was doing too, and with the ride hailing app still indicating that my conveyance was a mere 12 minutes away, for the last 30 minutes or so, I tried another service.
Second app
Having the second biggest ride hailing app on my device also, I tried it and found it to be little better than the first. Eventually, I used my smart phone in an increasingly uncommon, though very familiar function — I made a phone call, to a person.
I called a local cab company and asked of the possibility of getting a cab. I was told that there were cars in the area but that things were a bit chaotic due to the weekend that was n it and the big event in town, not only due to demand, but also traffic disruption.
Eventually, the staff said a taxi was available, and we were whisked off. The taxi was provided by the ride hailing app, the first, and so I cancelled my booking with the other. I also texted the local company to say we were sorted, because it was just good manners.
However, as we traversed the last roundabout before arriving in our own neighbourhood, I received a call from a taxi driver to say he was outside the venue, waiting for us. I inquired which service he was hailed from, and verified with the driver in front of me that they were from one and the same service and reported such.
To say the driver on the phone was irate is a bit of an understatement. He proceeded to berate me for a false booking and said he had travelled 12 km to fulfil the call. I apologised and said I had done all I could to prevent such an eventuality but that it was really down to the service to manage such things and hung up.
Voice analysis
Having used the service on other occasions and found it to be very good, it set me thinking as to why such things might occur.
Such services are built around smart apps and backend services, but it seems that whatever intelligence is built in to them, they could still benefit from a human touch at times.
I was fascinated a few years ago to learn that voice analysis in call centres had become sophisticated to the point where a caller, on first interacting with the system’s interactive voice response (IVR), would have various parameters of their voice measured and assessed, not just the nature of their request. One of the key reasons for this is that if a customer calls and is already highly agitated due to the nature of their query, then they will rarely be satisfied in being dealt with by an automated service, even if that service is attempting to connect them to the right resource.
If an irate customer can be directed to a human almost immediately, the result is that their query can be dealt with in a more personable manner that might well mean a higher satisfaction rate, and less irate customers overall.
Another such development is the around video analysis technology which has now developed to the point where it can be used for mass public events to sense mood and disposition to monitor for potential trouble spots before they become incidents.
These combined capabilities leverage the best of machine capability, where patterns are sought and identified in large amounts of data, to escalate them to humans for ultimate determination of the best way to rectify them.
Most of the time
The ride hailing app works very efficiently most of the time, but late on a holiday night, with demand rising and the patterns of usage straying beyond the norm, there seemed to be a breakdown in communication that not only left individual users wondering why the displayed information no longer reflected the actual situation, but also those paying a hefty subscription for professional use were left trying to deal with a set of irate proxy customers.
There was no support call, there were no escalation options, there was only an increasingly inaccurate supply of information that did not reflect what was turning up outside the door. The local cab company at least was able, via a human interaction, to convey more information for the context of what was happening than this scalable, cloud-based, globally available, service with built-in intelligence.
The point is that without that escalation capability, the situation got out of control resulting in dissatisfied punters, subscribers and drivers. Had the system had the capability to recognise that things were going pear-shaped and start escalating to humans to allow them to take some action to reassure the various stakeholders, it may have gone a long way to alleviating the situation.
As it was, the subscriber expressed an intention to look for alternatives, as did the few others I spoke to who had shared the experience.
Issues
Intelligence in apps is a good thing, and has led to improved user experiences in everything from billpaying, to food ordering and, yes, even taxi hailing. But when that intelligence runs into issues, it would appear a human layer is still not only necessary, but critical, as so much has been done to raise user experience expectations, such failures now appear in even more stark contrast.
Call centres and crowd control have provided a valuable model that might well be a better guide for B2C apps than is currently being applied.
Subscribers 0
Fans 0
Followers 0
Followers