Getting the right balance of human agents and automation is important for providing a good customer service experience.
Customer service "can be frustrating if not done properly," concedes LivePerson regional vice-president Steven Fitzjohn.
The move away from using the phone as the primary communications channel and towards SMS and various types of chat (eg, Messenger or Web chat) has the big advantage of removing the real-time element, thus taking a lot of the stress out of the conversation.
It also makes it simpler to automate the interaction, using bots to handle the high-volume, easy to answer enquiries, leaving human agents to deal with complex issues.
{loadposition stephen08}While bots have been in existence for around a decade, what Fitzjohn calls the "tango" between bots and human agents hasn't been done well until recently.
Bots can now do a good job of determining a customer's "intent" (asking for an account balance, advising a change of address, and so on), especially as one tends to use a relatively small vocabulary of around 10,000 words in this context.
They can also handle simpler transactions well. But what has proved difficult is the handover from bot to human agent. The technology is available, Fitzjohn told iTWire, but it has to be implemented well "otherwise it's a big waste of everyone's time" as a poor experience means the customer will avoid the automated process.
LivePerson finds it is generally better if a bot handles just one task. So a typical arrangement would be to have an intent bot that works out what the customer wants to do, and then hands over to either another bot that can perform the required task or to a human agent. (This approach is being used by LivePerson customer Royal Bank of Scotland.)
The advantage of messaging and chat systems is that the conversation continues in a single window, he said. There's no change of context for the customer, and the human agent can see the entire conversation.
Other bots may assist the agent, providing what it thinks is the best answer for the customer. The agent can either accept the answer or override it. This provides the customer with a good experience and helps tune the bots.
Whether the tuning is done implicitly by agents rejecting suggestions or explicitly, it should be done daily so the bots get smarter more quickly, Fitzjohn recommended, noting that this strategy is endorsed by many LivePerson customers.
And the most success comes from starting with a small number of processes handled by bots. Once they are working smoothly, more can be added over time. A big-bang introduction generally results in a poor customer experience, he suggested.
The combination of well-tuned bots and skilled agents does improve customer satisfaction. Bot-only systems have historically resulted in 60-65% satisfaction ratings, whereas the single-window blended approach is giving ratings of 80% and above.
One LivePerson customer in Japan used a bot just to handle password reset requests, with everything else being passed to a human. 90% of reset requests were handled by the bot (which is considered a good containment rate), and customer satisfaction reached 88.9% – "better than human assistance," according to Fitzjohn.
A recent survey commissioned by LivePerson found a majority of Australians have a positive or neutral perception of using a bot to communicate with a brand’s customer service. About 42% of Australian respondents rated their overall perception of bots as positive while 47% rated it as neutral.
Furthermore, 50% of Australian respondents believed bots are used to offer faster or better customer service as opposed to simply being a cost-saving tool for the company.
Yet 57% said they would prefer to wait a short period of time to speak to a human rather than chatting with a bot immediately. This seems to reflect a concern about how well the bot will perform, because in a scenario where a bot is just as accurate as a human customer care agent, 61% (and 68% of millennials) would prefer to chat with a bot over a human.