By Andrew Burgess
There are two opposing views about how software robots and artificial intelligence will impact the world of contact centers. On the one hand there is the pessimists’ view that robots will eventually take over all of our jobs no matter what we do, and on the other is the deniers’ staunch claim that robots will never be as good as humans in dealing directly with human customers. The truth actually lies somewhere between the two, and is much more optimistic than one might expect.
So, before we go any further, I just want to be clear what ‘robot’ technologies we are talking about. I am specifically focussing on two main types. Firstly there is Robotic Process Automation (RPA) which is clever software that mimics the work done by human beings to process work. RPA software is much cheaper to run than humans, and is much more accurate and reliable, and therefore represents huge competition to anyone that currently does a lot of transactional processing work, such as in a bank, utility or finance department. Secondly there is Artificial Intelligence which is even cleverer software that is able to make sense from unstructured data, such as customer emails. (RPA, on the other hand, requires structured inputs, such as forms). AI also has the ability to self-learn, so that it gets better at its job the more it does it.
Both of these technologies work in very different ways but the impact is potentially the same - the removal of work currently done by contact center workers. The reason I say that the truth lies between the pessimists and the deniers is down to the subtleties of which elements of work can be replaced. This is also the key as to why the optimists amongst us might be more on the right lines.
Multi-channel contact centers get their inputs in a variety of ways; this could be telephone, email, webforms, web chat, social media or even old-school letters. And in the hierarchy of contact quality, the telephone call rules. This channel is the one that gives the highest resolution rates and the best customer satisfaction, particularly if the query is non-standard or complex. And, of course, this is where humans excel - being able to understand the subtleties of what the customer actually needs (even if they ask for something different), the ability to calm down an angry customer or up-sell to a happy customer. Anything we can do to have the agents spend more time talking to customers must, therefore, be a good thing. Which is where the technology comes in.
AI technology now exists that can read customer emails and extract the relevant information to either direct the query to the right person to deal with or to pass it on to RPA robots for subsequent processing. Not only can the AI find the facts it can also work out the customer’s sentiments, even to the point of detecting sarcasm. A great example is with Virgin Trains. They use Celaton’s InStream to help categorize all the incoming email queries from their customers. So, if someone emails in saying, for example, that the WiFi wasn’t working on the 12:54 from Euston to Manchester, the AI can ‘understand’ the issue, extract all of the relevant information, and then route that query to the right person in the customer service organisation. Not only that, it will also check to see if other people have complained about the same thing, whether that person has complained before and whether they are a VIP of some sort. So, when a human agent gets the information, they have everything they needs to process the query. And now, Virgin Trains have just started to automatically refund customers for late running trains, so the whole process, from the customer writing in to when they receive their compensation, can be handled without it touching a human being.
There are also a raft of AI ‘avatars’ that one can chat to via the web or intranet. In many instances, because the Natural Language Processing is so good, it is difficult for customers to discern whether they are chatting to a real person or a computer. The most famous example of this type of software is IPSoft’s Amelia which, to date, has been deployed mainly in internal help desk environments - it works perfectly alongside an existing team because it continually learns from them. Any incoming query it can’t answer with confidence it will escalate to a human expert who will, in the first instance, answer the query on behalf of Amelia. But the important thing is that Amelia then learns the answer and will be able to sort that problem on its own next time it appears. This then frees up the human staff to focus on the most complex queries.
At the back end of the workflow, actually processing whatever it was the customer rang up or emailed in about, the RPA robots can take on a lot of the heavy lifting. Whereas an agent might have had to spend valuable minutes wrapping up the call, keying in additional data or sending messages to other departments, the majority of that work can now be automated using RPA software, leaving the agent to get on with talking to more customers. Implementing RPA is very different to implementing any other IT system - the software is ‘trained’ to do the task exactly as the human would do it, so there is no disruption to the underlying systems - everything works at the presentation layer. And, of course, the robots can work 24 hours a day, 7 days a week if necessary, which is why the cost of running a robot can be around a ninth of the cost of employing a human being.
As you can now hopefully work out, a combination of AI and RPA can be very beneficial to the contact center environment. As in the Virgins Trains ‘Delay / Repay’ example above, for a large majority of the non-voice queries an AI / RPA blend can provide human-free processing, which will be faster, cheaper and more accurate. And, of course, leaves the human agents to spend more time talking to customers.
Any contact center should therefore be looking very seriously at the benefits that both RPA and AI can bring to their operations, both from a cost perspective, but, more importantly, from the freeing up of time it provides to the agents. This can only mean higher customer satisfaction as their simple queries are dealt with efficiently whilst the trickier ones are given the care and attention they require.
There will always be the pessimists and the deniers in the contact center world, especially around technology, but if the customers are clearly happier then, in my view, the optimists will undoubtedly win that particular argument.