Chatbots – The future of CX? Part I

Chatbots have surged in popularity, but will they stand the test of time?

Chatbots have surged in popularity, but will they stand the test of time?

Chatbots are a quickly emerging form of customer service which utilise AI to communicate with consumers but how useful are they really? This first part of our series will discuss what they are, how they work and ultimately what that means for businesses.

Developments in internet connectivity and smartphones have led to an evolution in how we communicate. Phone calls are no longer the easiest way to receive information. Text, image and video messaging apps, such as WhatsApp, have become the norm to reach out to friends and family. In 2015 messaging apps surpassed social media apps in usage and this growth has continued ever since. In the wake of this change an opportunity has presented itself to businesses. Live chat and chatbots are beginning to emerge as channels for brands to reach out to consumers.

But what is a chatbot?

Essentially, a chatbot is any programme or technology created to simulate a human conversation. This definition stretches from Alaska Air’s Jenn, one of the earliest commercial chatbots from 2008, to the much more recent and widely known Alexa from Amazon.

Alexa can recognise speech and will accordingly mimic human interaction to answer in a conversational way. Alexa can tell you about the weather, nearby shopping options and can even complete tasks like checking your credit card balance, paying bills or ordering an Uber. For this to work, Alexa must understand requests in order to answer them.

How do they work?

Alexa works using two main technologies – natural language processing (NLP) and speech recognition. NLP translates language into its core meanings and is necessary for any kind of chatbot to contextualise written or verbal data.

Usually, sentences are simplified and broken down, words are categorised as nouns, verbs or adjectives before relationships between words are found. Words that have similar meanings and sentiments are then recognised; happy, happily and ecstatic will all be linked together.

And it doesn’t end there. Word order, similarity to other words, isolated words, combinations of adjacent words and grammar are all considered to reach the goal of understanding sentiment.

I would like my new phone if it had an audio jack.” and “I like my new phone, the design is great.”

A rudimentary chatbot may examine each word in isolation to pinpoint the keyword “like” in both sentences. This would then lead the chatbot to determine that both statements are positive. It’s only when conditional tense and word combinations together are considered that the first statement is seen as negative.

Good NLP is not the only factor a chatbot needs to understand this difference. Chatbots must be “trained”: exposed to vast numbers of discussions and conversations to see how humans react to different interactions. The larger the volume of discussions or conversations the more accurate and realistic the chatbot becomes as it will be able to mimic responses for different sentiments and statements. This is paramount as it increases the context given to chatbots, even the most complex chatbot cannot identify a negative statement if it has never encountered one.

Speech recognition is packaged around NLP to translate audio patterns and verbal communication into a format that NLP can examine and break down, this then allows Alexa to understand audio patterns and verbal communication.

Why are they so useful?

To become the future of customer experience chatbots must succeed on two different levels. They need to provide enough utility to customers while providing ample benefits to businesses.

For customers, chatbots have one incredibly powerful selling point. Instant responses. At any time of day, a chatbot can provide a response to questions, fulfilling the growing demand for immediacy in shopping, communication and even lifestyle that has been developing from the rise of messaging apps. 69% of people cited instantaneous responses as their main deciding factor in reaching out to a chatbot before a human. Additionally, as chatbots can be created within existing messaging apps and websites it means that downloading new apps is not a requirement. Increasing ease and flexibility of communication will allow consumers to interact with businesses more freely, which alone will improve customer experience.

For businesses, chatbot response time provides the benefit of being incredibly efficient. They process large volumes of requests at once, lowering necessary manpower and improving cost efficiency. They can supplement human interaction while acting as a gateway for customers. Simple issues can be dealt with immediately while complex issues are piped to the right departments to be resolved. Chatbots already have large projected savings in banking, healthcare and retail – where they are estimated to save 2.5 billion man-hours, equating to US$11BN, globally by 2023.There is a strong financial argument to adopt chatbots into the customer experience arsenal of any business.

Lastly, a factor that works favourably for both customers and businesses is that some chatbots continually learn. Through numerous conversations chatbots hone their responses, helping customers more efficiently over time while providing businesses with a wealth of data on customer preferences.

In this post we have outlined chatbots and their merit in customer experience, however, this is just one side of the coin. In the next part of the series we discuss chatbot weaknesses, and their current and future development paths.