Chatbots: The Smart Decision on Customer Service and Artificial Additives

By: Dale Roberts

March 7, 2017

Clarabridge Analytics
Clarabridge Engage
Contact Center
Customer Experience
Text Analytics
Artificial Intelligence
Machine Learning
Speech Analytics

Earlier this year, I found myself in a long queue of unhappy travelers. My 1am flight from Bahrain to Heathrow had been cancelled shortly before departure. I had a long night in the transfer lounge with my many travel companions, as a very limited number of customer service agents tried to find places for a truckload (or more accurately a planeload) of passengers.

Much of the anxiety of my unplanned, overnight, and unscheduled stay in Bahrain could have been alleviated by the most basic of algorithms offering me the next flight, with any carrier, back to London.


Meanwhile Silicon Valley and its UK cousin, Silicon Roundabout, are constantly cooking up ways for software to move customer experience out of the Dark Ages. The latest digital wave is small, discrete pockets of artificial intelligence bundled up into a friendly assistant. These will help you re-book a flight, buy a pair of shoes, or shop from Facebook Messenger, Skype, or Slack without any human interaction.

Chatbots are a fusion of machine learning and natural language processing which are starting to be a factor in customer service. Today’s chatbots include Operator from the founders of Uber, a personal assistant for those with busy calendars, Royal Bank of Scotland’s Luvo in the financial services industry, and beyond.


Right now, the hype around chatbots exceeds the reality. This should come as no surprise. Who hasn’t had at least one maddening exchange with Siri, Google Now or Cortana? All you want to do is send a simple message, play a song or check a movie start time. But your assistant can’t seem to keep up. Really Siri, you can’t find any good music? It’s clear that we’re in the early development stages still.

When Mat Honan of Buzzfeed experimented with M, a concierge chatbot in development from Facebook, he discovered that human workers completed his requests. Like The Wizard of Oz or Mechanical Turk, there was thunder and lightning up front, but human button pushing behind the curtain. Some critics have suggested that as many as 7 out of 10 requests require human intervention. Microsoft have not been doing much better in the AI race. They released a replacement, Zo, after online tricksters taught their original chatbot, Tay, hateful, racist language.


The challenge is not just digital teething — it is something more fundamental. Donna Peeples, Chief Customer Officer for Pypestream, compares the current state of chatbots with the early stages of IVR systems. Peeples says “You’re given a set of choices and you are not interacting with a human. At its most basic level, it’s a digital IVR.”

This is not good news for customer experience, and Peeples would be the first to agree. US industry group Interactions asked 1,300 customers online earlier this year if they liked IVR. They found only 3 percent said they did. I can only hope the cost savings justified a technology that 97 out of 100 customers did not like.


So Chatbots have to do better than IVR. Much better. Luckily, recent innovation and successful use cases are encouraging. Lufthansa recently announced a service that responds to inquiries about flight status, baggage allowances and lounge access through Facebook Messenger in a matter of seconds, 24 hours a day. Chatbots don’t sleep, which gives them a leg up on man-powered customer service.

Meanwhile, Mastercard intends to launch a service next year that will check account balances, dispute credit card transactions or pay a bill, also through Facebook Messenger. In essence, we are seeing a set of services emerge that are served by chatbots that communicate with us conversationally more efficiently and on an on-demand basis. No lines, no waiting.

However, the use cases are somewhat limited and could easily run into problems. It’s all too easy for the customer service interaction to become more complex than the decision tree of responses that even the most sophisticated of chatbots can handle. What if a customer has multiple orders, is travelling with a companion that paid for them both, or is using their company account and not their personal one?

The mechanical turk points us towards a solution here. The secret is to enhance, not replace agents, to help customers resolve issues faster and reduce wait times. However, organizations must orchestrate the integration between chatbot and human agents perfectly. The number one frustration for customers, according to Harvard Business Review, is when they are asked to repeat themselves. The chatbot might be a model of natural language and AI genius. However, if the handoff is slow or incomplete, then the customer experience will implode.


Thankfully, we can learn a lot from our previous attempts to automate customer service. In fact, it might be the only good thing to come from IVR, which was all too often justified by call deflection and reduced spend. It was an approach that made it low cost and efficient to deliver customer dissatisfaction. Well done, us.

Instead, we should let customer experience, not cost, drive the implementation of chatbots. Enlightened brands are taking the approach that they can enhance an agent with support from chatbots when a rapid, routine, repeatable interaction can be driven more efficiently for the customer. This blended approach could prove to be far more successful. Customers can spend less time in line and agents can spend less time on the repeatable interactions that very quickly get old for them.

The chatbot revolution is very much a work in progress. However, there is already a smart decision tree that should be the guiding light for every business’s approach to implementing them. It looks like this:


10         THEN




50         ENDIF


70         THEN

80         GOTO 00

90         ENDIF



Dale Roberts is VP of Professional Services for Clarabridge, author, commentator, columnist, and speaker. As a professional services leader for Clarabridge in Europe, Roberts is advising some of the world’s largest companies on optimizing the customer experience using social and digital insights. Prior to this he was part of the founding circle of Artesian Solutions, an innovator in social CRM and a Director of Services for business intelligence giant Cognos. His first book, Decision Sourcing, is an inspiring commentary on the impact of social on corporate decision making. His latest, World of Workcraft, is a timely piece on engagement, motivation and digital humanism in the workplace.

Dale is one of the top thought leaders in big data and analytics by Analytics Week, a contributor to business and technology publications including Wired and ClickZ and a Fellow of the Royal Society of Arts. Follow Dale on Twitter: @DaleSRoberts.


Clarabridge is at the cutting edge of customer experience management. Request a demo today.