Organisations are using conversational AI to relieve the pressure on their contact centres while at the same time aiming to improve productivity and service levels. In the past, the performance of chatbots has been somewhat poor and the results disappointing. In recent years, however, AI technology has come a long way and some organisations at least have started to mature in their understanding of its potential use and application.
When we talk about Conversation AI the focus tends to be on chatbots and virtual assistants, but as Tim Sheedy, Principal Adviser at research and analyst firm Ecosystm, points out, it embraces much more. He says, “The concept of conversational AI is bigger than just chatbots. There is voice intelligence and video intelligence, where the AI is able to understand emotional responses from facial expressions and tone of voice. So, if someone calls the contact centre and they sound distressed they will receive a different sort of service than someone who doesn’t sound addressed.”
Techtarget defines conversational AI as a type of artificial intelligence that enables consumers to interact with computer applications the way they would with other humans. Erica Johnson, Solutions Director at Curious Thing, elaborates, “Conversational AI is anything that allows a machine to talk back to a human where the machine is smart enough to understand the context of what’s happening. AI can be applied to any kind of conversation that you can imagine. There’s the traditional customer service type of things like answering questions, but we’re seeing interesting changes where the AI is starting to be used for conversations like counselling and support”.
Conversational AI allows artificial intelligence to gather information, resolve problems and enquiries, execute transactions, provide after-hours support and streamline service. Natural language processing (NLP) and machine learning (ML) allow chatbots and virtual employees to recognise human text and speech. Johnson has seen this technology advance rapidly over recent years, “The tech has got a lot smarter over the years whether it’s the speech to text or the text to speech components or the brains that are being built to understand human conversation and generate responses. I think there’s been a natural evolution and it will continue to get better.”
Exponential growth in conversational AI
In the next few years, it is predicted that investment in the technology that enables these conversations will grow exponentially and the impact on the contact centre environment and CX teams will be dramatic. Globally, according to Grand View Research, the conversational AI market is expected to be worth USD 41.39 billion by 2039. The market growth is being fuelled by the rising adoption of advanced AI technologies where companies are viewing it as a means to reduce costs while accelerating growth.
Gartner said that using conversational artificial intelligence (AI) in contact centres could cut labour costs by up to $80 billion in 2026. Daniel O’Connell, a VP analyst at Gartner, said that AI could help companies challenged by staff shortages and poor customer experience. However, Gartner’s report stated that the fragmented vendor landscape, complexity of deployments, and integration costs could result in lower adoption over the next two years.
Focus on the customer experience
Investing in conversational AI primarily as a means to reduce pressure on the contact centre and reduce costs, however, is likely to lead to failure. Sheedy has observed, “Companies will use a chatbot to take a bit of pressure off the contact centre and to give customers quicker access to answers. The data supports the idea that if a customer goes to a chatbot and they don’t have their questions answered and then have to speak to the contact centre, they will be less satisfied and more frustrated than if they called the contact centre in the first place”.
Rather than using conversational AI to reduce costs or raise productivity, Sheedy recommends companies approach it as a means to improve customer experience. “The companies that are doing it well are the ones that first have invested more in the AI in terms of understanding the customer journey and the customer intent. The AI is smart enough to understand what the customer is trying to achieve and their intent rather than just recognising what they are saying”.
Erica Johnson believes that many of the bad experiences with conversational AI and chatbots in the past are due to the chatbot or virtual agent being given work that humans don’t want to do; work that’s deemed of low or minimal value. She says, “I believe conversational AI is successful when it is given good work that is of value to the customer and the business, but it’s repeatable and predictable. The AI can be trained to do a really good job while the humans are off figuring out the things which are not yet so predictable and have more volatility in it”.
Successful chatbots allow people to get something done or achieve an objective.
Successful chatbots allow people to get something done or achieve an objective. A large percentage of chatbots, however, are designed simply to provide customers with information or access to a knowledge base of articles. It’s then up to the customer to decide what to do with that information and navigate their own path to resolve their enquiry. Sheedy says, “Let’s say we have a chatbot that, based on a query for a mobile phone, presents the customer with a series of plans. When the customer selects the plan they want, they are then transferred to the contact centre. Instead of this, let’s have the chatbot actually guide the customer through the process of signing them up to the plan.”
Sheedy highlights the implementation of CommBank’s chatbot as a good example of getting it right, “The CommBanks chatbot not only handles 78% of the bank’s enquiries, but it also resolves them without having to transfer the customer to the contact centre. During the pandemic, when people were ringing up distressed about their home loan situation, the chatbot went straight to the point of helping them with relief from their home loan payments for a set time. The chatbot actually does something versus just giving people information”.
When and where to use conversational AI
Conversational AI is applicable in a broad range of industries for a variety of tasks. One of the main areas that it can be useful is in after-hours customer support. Customers want real-time support that allows their enquiries to be resolved quickly, and robots or virtual agents can provide that support 24 x 7, seven days a week. Johnson says, “Some customers for a variety of reasons can only deal with the business after hours. So, what do you do to provide those people with support? Do you pay for an after-hours concierge type of service or do you try and get off-shore agents to work those hours? What is the most cost-effective means to deliver after-hours support?”
“There’s so much you can do with AI to guide people through self-service, handling transactions and accessing information from the CRM to answer simple enquiries. Another useful application is rich information gathering. These involve very structured conversations that involve gathering rich information such as references for recruitment purposes or conducting surveys for market research.”
Learning from past mistakes
In the past, the performance of chatbots has been somewhat mixed. Even with implementations that haven’t failed outright the experience customers have had with chatbots has not left a positive impression. Chatbots have developed a reputation for providing poor and efficient service. Sheedy notes, “Unfortunately, despite some notable successes, there are still plenty of chatbots out there that don’t really understand the customer intent, don’t have that much intelligence, have a very limited set of questions that they can answer. A lot don’t even answer questions, they direct the customer to read this document or go to this web page”.
Developing a chatbot without a clear idea of the customer’s intent and the stage they’re at in their journey is unlikely to lead to success. Johnson mentions how organisations have tried to use AI to automate processes without fixing those processes in the first place. This approach just creates more problems and frustration for customers. She also highlights that many chatbots have been released with really limited information and capability. “If you’re releasing something to service a customer problem it has to be able to serve the customer problem. In the past, many chatbots were released with very limited information. But you need a very understanding customer base to get away with that and you need to move very quickly to make sure the machine learning is adapting as you go.”
Conversational AI offers a lot of opportunities. It is an exciting area of development for customer experience and digital technology where its application is wide and varied. But there are also significant challenges which need to be navigated. Sheedy warns of an emerging gap in the Australian market between companies that will invest wisely in conversational AI as well as CX and digital transformation and those who will fall behind. “There’s data emerging that shows that there’s a real gap opening up in the market in Australia, between the haves and have-nots when it comes to customer experience and digital experience in particular. There are those that are investing and making it better while others are holding back budget for a number of reasons.”