Lee Lofthouse

e-commerce Business Consultant

The great chatbot paradox


Chatbots - the future of customer services, yet business knows that customers don’t trust them.

There does seem to be little doubt that if a well implemented and relatively intelligent chatbot can deal effectively with customers’ FAQs - the bread and butter of these AI enabled programs - then real people can deal with more complex problems, or those requiring the human interaction which will inevitably be needed when customer issues are escalated. Here businesses can ultimately reduce cost and improve customer experience - or so the saying goes.

Unfortunately, it isn’t quite as simple as that - the vast majority of customers simply don’t trust chatbots because they don’t believe that they are smart enough to help them solve their problems.

The chart below from MarketingCharts.com clearly demonstrates much of the skepticism surrounding the use of AI in customer services. It all boils down to the fact that a resounding 9 out of 10 businesses and consumers say that customers still prefer to to be served by human beings.

Source: MarketingCharts

Unless an AI enabled customer service program undergoes a process of continual improvement, with the ongoing fine tuning of its accuracy, user experience will suffer. Ironically, we then risk ending up in a scenario where customers start trying to be understood by the machine, which will take people straight back into the arms of real life customer service agents soon enough.

There was an excellent quote from Colin Allen in an article for CIO titled ‘How chatbots like Siri will get smarter’, which totally encapsulates the dilemma:

"We need to be more attentive to issues of how humans adapt to inflexibility (and inscrutability) in the machines. We make assumptions, often false, about their capacities, and then adjust our behavior to their inflexibility".

Colin Allen

Distinguished Professor

University of Pittsburgh

A good example of best practice is in the banking sector, which represents one of the biggest challenges in terms of the complexity of interactions with customers. Here clients place very high expectations on their bank to provide a smooth and highly efficient user experience, which is humanized, and yet not human. According to Ilker Koksal writing for Forbes in his article entitled ‘The 9 Steps to Build a Great Banking Virtual Assistant’, “conversational banking is the new digital framework of banking”.

If we stick to the role of such bots in a ‘customer-facing’ sense, a VA uses something called ‘dynamic conversation flow’ whereby the program not only recognizes what the user is asking and knows where to find that information, but is also able to generate a predictive natural language response. The benefit is that it can present information in a more dynamic and reactive way. Typical chatbots lack this flexibility, and can -for example- easily lose track if a customer uses less predictable language, or goes off topic, which leads to a breakdown in the conversation.

As the market moves slowly away from the very mechanical feeling of early chatbots, the key is to give customers the instant service which is increasingly expected, without needing to scale up customer service centers to unsustainable levels. Chatbots offer really significant cost savings and should bring a consistency which would be hard for human beings to beat, provided vendors are fully engaged with the ongoing monitoring of performance.