By Scott Logie, MD, Insight at REaD Group
The robots are coming! Well, not quite, but the ever growing trend for implementing AI and automated systems to aid in our everyday lives seems to be showing no signs of slowing.
Recent research conducted by advisor company, Gartner, suggests that by the year 2020 a quarter of all customer service and support will incorporate chatbots or a similar form of virtual customer assistant technology. This seems like an astonishing figure, and one that has both positive and negative connotations.
The last decade has seen a huge jump in the proportion of people engaging on digital channels, and it is therefore hardly surprising that companies are investing more and more in these virtual customer assistants. Purely from a resource point of view such a transition makes a lot of sense; artificial intelligence (at least at this stage!) doesn’t ask for a pay cheque.
There is also a distinct advantage to the consumer – these automated systems are capable of functioning 24/7, without the need for sleep or coffee breaks. Furthermore, the prospect of a future free from hours spent on hold listening to Justin Bieber might not be the worst thing in the world.
However, when it comes to customer service, can human contact ever truly be replaced or replicated? According to research we conducted last year, 62% of consumers rated high quality customer service as the largest factor influencing brand trust and loyalty in the retail sector (Retail Trend Report 2017: New World, New Consumer).
While these virtual customer assistants are undeniably becoming more sophisticated all the time, it is the warmth of human interaction that creates this customer engagement. Some of these systems are capable of detecting frustration and anger in a customer’s voice and will transfer the call to a person in a call centre at a certain point. I find shouting down the phone helps. But by and large there is no doubt that they are still worlds away from being able to react and alter their response or attitude based on things like sarcasm and emotion.
There also comes a stage when we have to ask – where does this end? Do we eventually reach a point where human contact has been phased out entirely and we find ourselves reliant on machine to machine relationships? Say, for example, my bank bot detects that I’m overdrawn and applies for an overdraft on my behalf, and this instigates another chatbot which then decides whether to grant me said overdraft. The possibilities are dizzying, and somewhat terrifying – just one short leap to Skynet!
The question of trust and customer experience is not one to be overlooked lightly. The majority of consumers taking part in Gartner’s survey said that they find it difficult to trust VA’s to assist them with more complex tasks, such as handling their banking, insurance or utility issues (29%, 16% and 35% respectively). Therefore, brands will need to demonstrate to customers that they will still receive the same high levels of customer service once these technologies have been incorporated.
Perhaps if this predicted future comes to pass a balance will need to be struck between convenience and functionality. A system whereby technology and human work in tandem may be considered as an initial compromise – HAVA’s (Human Assisted Virtual Assistants). The idea being that when a VA is faced with a situation it cannot handle or a question it cannot answer, a human agent will then take over the conversation. The growing development of machine learning will also theoretically mean that VA’s are able to learn from these instances and adapt to resolve these situations themselves in future.
Essentially, it will remain to be seen how effective these VA’s are in maintaining the high levels of customer service that consumers have come to expect. Advancements in technology such as natural language-processing and machine learning are perhaps bridging the gap between the soulless and robotic automated systems that we’ve come to know, but can they ever truly hope to encourage the same level of engagement and replace human interaction? The next few years will certainly be interesting, but brands must be sure to put customer experience first, or risk dealing with the consequences.
2nd February 2017
By Scott Logie, MD, Insight at REaD Group
During the recent, and really rather excellent, DMA event on the Future of Customer Engagement there was a throw away remark made by one of the speakers that really stood out for me. In amongst all the Dystopian nightmare futures; predictions of millions of job losses and the creation of new ones we can’t even imagine as yet; the replacement of call centres with chatbots and the running of our homes handed over to tiny lovable robots was the statement “one aspect of customer engagement we need to consider is that much of it will be between machine and machine”.
When we think about customer engagement, at least when I do, we tend to think about computers, or even Artificial Intelligence on the company side – creating triggers when someone drops onto a website or automating outbound comms, responding to live chat on-line or in a call centre. However, there is also the automation, or computerisation, of the customer side of things. With things like Siri, Cortana, Amazon’s Echo and Dot functionality and Google Now already common place there are times when the decision to engage with a website or brand may not be made by the end consumer but could well be made by a “machine”. As such, who – or indeed what – is the customer in this instance?
From a data perspective, this makes how we gather, hold, manage and then present data and information in the future very interesting. In the past a Single Customer View (SCV) was centred around a person, or maybe a household, based at a physical address. Over time this evolved to take into account e-mail address where we would often know an individual by an email address, and start to gather a lot of data against that email address without knowing who the person was, or where they lived. In many cases it might even be that we never actually find out what that person’s full name is but through e-mail we can build up a decent amount of knowledge and information about them and start to create a meaningful relationship.
In recent times that has evolved again. From the first time an individual drops onto a website we now tend to capture the IP address and/or the device ID of the device being used. If that doesn’t connect to one we already know we then create a new record in the SCV. Over time this device connects hopefully to an email address as we get initial registration and then, if needed, to a name and address as a transaction takes place. This does mean that we do already centre our SCV developments on machines.
To a certain extent it is about ensuring that we capture the most relevant data as early as possible and then connecting that data to a “person” as we learn more. So in many ways we have the technology capability already in place to deal with building a database, or single customer view, that creates customer engagement strategies between two devices. On one side are automated triggers and engagement streams already in place, on the other side devices that may or may not be operated by a human.
Back to the recent event. During an inspiring presentation by Jeremy Waite from IBM, he introduced us to Jibo, the next generation of home help style AI. Jibo responds to human instructions but also has enough intelligence to collate data and make decisions on its own behalf. As an example, Jeremy, who has been testing it out, recently got told by Jibo that he was running late for a meeting and didn’t have time to walk so Jibo had ordered him an Uber. While this is amusing and shows how AI is developing, it also raises quite an important question: Who are we marketing to here? The person who owns the robot or the robot itself? Does Jibo decide to use an Uber because the owner has trained it to select Uber as their taxi firm of choice or does Jibo select Uber because it recognises Uber from marketing messages received? To a certain extent it doesn’t matter, as AI develops the robots will increasingly make decisions on our behalf, marketing will need to also develop to “sell” to machines.
While we might be set to market machine to machine at the moment, and be able to gather and host the data required to do so, I suspect where we currently do fall short is in our ability to distinguish between machine and person in terms of how we are trying to influence and why. This will provide a real challenge for the future and one that might need to be solved quicker than we think.
Want to talk about the power of insight ? Contact us today