Every year companies spend $100+ billion annually in wages to call center reps. As a consultant, I have seen numerous companies where most of customer service queries were about issues that could be resolved in self service channels. Then came chatbots! However, the world of customer service has changed only marginally. We explore the reasons behind that.
Chatbot / conversational AI development is accelerating
Now companies have a unique channel which can both serve simple queries at low cost and serve complex queries by seamlessly transferring customers to a human agent. On top, bots are lightning fast, one of top criteria for customers. As a result, search queries on chatbots quadrupled between 2015-17 and investment also increased. Tech giants built an ecosystem of bot analytics and development tools.
Then why am I still talking to customer service reps? And why has interested in chatbots remained stable from 2018 to 2021? I worked for about a decade as a consultant and implemented numerous process automation projects. I see these main obstacles: Discoverability, customer expectations and technology limitations. Given that today we rarely talk to bots, these obstacles remain.
Discoverability: There is still no single place to discover bots
I have an urgent need as a customer and need to access customer service. What do I do? There’s a myriad of options which may or may not result in a customer service interaction: go to the company website hoping that they have a chatbot there, download the app which may not even exist, check out whether they have a chat bot on a platform like Messenger. Most Fortune 500 companies do not have a chatbot on any of these places. So the current situation is bleak in the US. Enter WeChat, dominant chat platform in China where virtually every major corporation offers a text interface for customer service. A dominant platform can make all the difference.
Even a system like Amazon’s Alexa with continuously expanding skills (currently it has >100K skills) has discoverability issues. How do I know which useful skill Alexa gained among the thousands of skills?
Hype: Customers expecting to talk to smart bots, find that they are talking to Office Assistant
Most chatbots in WeChat are essentially handcoded FAQ services. With news on the Western media about robots replacing hundreds of thousands of workers in the next decade, customers have understandably high expectations from a chatbot. Facing an interface akin to 1970’s Teletext, which works well in WeChat, can be crushing.
Finally, technological limitations remain. An AI interface that can solve all of customers’ problems without human interventions does not yet exist for any business domain. This is because language understanding is a really hard problem that may take decades more to solve at a human level.
We outlined the challenges but we should also highlight the opportunities! There’s a vibrant chatbot eco-system and chatbots are here to stay! Our next piece was regarding strategies on how to overcome all of these challenges so your chatbot can have a chance of success!
 Total annual pay of customer service reps in the United States in 2014 assuming median wage equals average age. Source data from Bureau of Labor Statistics. Taking into account 2017-2014 growth and US companies’ spending to outsource customer service, by 2017 this should be $100+ bn in 2017.
How can we do better?
Your feedback is valuable. We will do our best to improve our work based on it.