Over the last ten years, banks have transformed the way they serve customers. They’ve built mobile apps, improved their websites, and launched chatbots – all of this so their customers can access information and make transactions 24/7. But despite their efforts, banks still have large call centres receiving huge numbers of calls. The Covid pandemic has further increased the strain on already stretched call centre resources.
“The super majority of banking conversations with customers
are ripe for automation.“
It is expensive for a bank to talk with a customer, even if the customer’s inquiry is as simple as checking a recent transaction. There are a number of factors that determine the cost – and of course the duration – of a call. Whereas some calls may cost just a few pounds, others may be £30 or more. The sting of this is that many calls are focused on relatively simple and everyday banking housekeeping – moving money, balance queries, checking on a transaction, or finding out whether funds have been sent or received. The upshot is that the supermajority of banking conversations with customers are ripe for automation.
Banks are not unaware of the challenges involved in simplifying the way we do banking, and continually present new capabilities via their digital channels. However, as an executive at a large UK bank said recently, “every time we put something new in our mobile app, our call volumes go up. We just can’t win”. Customers either can’t find the newly launched feature, don’t know how to use it, or want reassurance that it will work. And so customers take the comfortable and natural route and seek to talk to someone – a step that typically requires expensive calls with contact centre agents or visits to bank branches. Indeed, someone has even suggested that if you want to rob a bank, just keep calling their customer service number.
The current situation isn’t good for customers either – long wait times and the inconvenience of making a call are all reasons to put off dealing with non-urgent bank transactions. “I’m really looking forward to calling my bank today” isn’t a sentence you’ll often hear in the real world.
“Customers need convenience and efficiency but they also
want to be understood first time, every time.”
Chatbots have been presented as a solution to the customer service headache – and in some cases, offered as a panacea. Chatbots enable a customer to engage in a conversation with an automaton (almost always via text) at any time of day or night with the goal of replacing live webchat with customer services. Customers benefit from being attended to quickly – and at times experiencing an efficient and supportive process. The other big claim is that chatbots can reduce costs for banks.
In terms of chatbots, NatWest in the UK has Cora, HSBC has Amy, Bank of America has Erica – and other international examples have a similar response to the customer service conundrum. These chatbots have utility and deflect a minority of customers away from calling contact centres. However, they are not the be all and end all. They often fall short of most customers’ expectations as well as those of the banks that operate them.
It has proven to be much harder than expected to automate the majority of conversations. This is not because the banks are inept, but simply because human language is so complex and chatbots simply cannot handle the everyday back and forth of natural conversation. When chatbots get it wrong, they tend to get it totally wrong.
Some startups have entered the fray alongside the banks. Cleo, for example, offers a service to customers that enables them to monitor their spending and interrogate their bank accounts. Others have helped banks launch chatbots to their customers.
Some of these initiatives have a positive impact, but the real prize is being able to automate interactions and support rich queries over the phone and at scale. To prevent ‘the bank robbery’ of customer service calls, banks must deliver delightful automation that delivers all the benefits that chatbots promised – but over the phone. Customers need convenience and efficiency but they also want to be understood first time, every time. And people already know how to make a call and how to interact over a call so they’re not being asked to change their behaviour.
“We now have the capability to automate customer service calls
over the phone in a way that has previously been impossible.”
Today’s next-generation automated assistant doesn’t need to be spoken to in any particular way, so once on the end of a line with an automaton, we shouldn’t need to be “more robot” in order to be understood. The automaton should be human-like in its understanding when we talk to it naturally. This doesn’t require magic or mind-reading, just very deep capabilities in conversational AI and an ability to process information in context. If a customer says something within the automaton’s remit, they should be understood first time and their query resolved.
With new breakthroughs in machine learning, action.ai now has the capability to automate customer service calls over the phone in a way that has previously been impossible. The vast majority of calls can be automated to deliver delightful customer experiences and encourage customers to take greater interest in the services offered by their banks. action.ai can now ensure that it is no longer possible to rob a bank by simply calling them – the calls will be too cheap and too effective.
