Gartner: Why humans will still be at the core of great CX
- 20 June, 2018 07:05
Companies looking to automate all their customer experience using artificial intelligence (AI) or machine learning are likely to fail at delivering the seamless, post-channel CX experiences customers want.
This is the opinion of Gartner VP distinguished analyst, Michael Maoz, who said it is not a matter of bots/AI replacing all human interactions in future, but actually facilitating humans to deliver better CX.
Speaking at the recent Gartner Summit in Sydney, Maoz noted 1600 startups in AI this year with $12 billion in private equity/VC investment. By 2022, the analyst firm anticipates 72 per cent of customer interactions will involve an emerging technology, such as a machine learning (ML) application, chatbot or mobile messaging, up from 11 per cent last year (2017). By 2021, 15 per cent of customer service interactions are expected to be completely handled by AI, an increase of 400 per cent from 2017.
Phone-based communication will drop from 41 per cent to 12 per cent of overall customer service interactions over this period. However, a human agent will still be involved in 44 per cent of all interactions, and by 2020, 40 per cent of bot/virtual assistant applications launched this year will be been abandoned.
In 2017, customer self-service sat at 48 per cent of all interactions, and is expected to grow to 64 per cent by 2022. Meanwhile, agent-based interaction will reduce to 15 per cent by 2022, with a further 21 per cent being agent-assisted self-service.
Maoz told CMO AI should be used to help with non-complex tasks, therefore freeing up human agents to build relationships. AI will also move conversations ‘post channel’, or beyond omnichannel and into a singular whole customer experience, he said.
“Conversations will be AI-guided instead. Where the process is transactional in nature, rather than relational in nature, AI can shine. If it’s binary and very clear, we can use AI to sort it out,” Maoz explained.
“However, I can’t use AI to sort out more complicated questions, because that would involve critical thinking or symbolic analysis. That is an area where you need to speak to a human. These are questions where there is no right or wrong answer.
“If I’m also trying to build a relationship, and I don’t have enough data, a human can reach out and build that relationship. I could let you do it yourself, but for the relationship, I would prefer to bring in a human to grow the business.”
If organisations intermediate through AI, however, it’s possible to see what a customer has been doing then reach out. Maoz suggested AI is more likely to be a bad experience in this scenario, adding it’s easier to be ‘close’ human-to-human.
To identify where a distinction should be made, Maoz advised organisations to have fail-safes.
“If you’re running a chatbot, set the threshold for accuracy, so a human can look at the supplied answer and enrich it, and then get back to the customer,” he said. “Then, you’ve got a dialogue going. In this way, content is passed to a human and the human takes over.
“Now, whether a customer knows the interaction is with a human or not, they have the feeling of accuracy.”
The best AI applications today are the one that have these fail-safes, Maoz said. “If the AI doesn’t have the fail-safe of the human, or if a human doesn’t have all the information they need from the AI, then it becomes a bad experience for the customer.
“The idea is to make it seamless across channels, then the idea of multichannel or omnichannel disappears, it’s post-channel - it’s just one experience,” he added.
Gartner other recommendations are to redefine the mix of channels that will best respond to emerging customer desire for highly intelligent self-service. It’s also important to state clearly where human support agents will make a significant difference over automated response and advice.
Finally, select areas where advanced analytics will be used to collect insights and guide the business on how to better/more profitably service customers.