Are Chatbots Capable of Empathy?
There is a common perception that chatbots struggle to display empathy. But then, who truly excels at it? One of my guilty pleasures is reading the many consumer complaints featured in weekend newspapers—yes, the traditional printed editions that get ink all over your hands for our younger subscribers. And the reasons aren’t to take joy at other people’s discomfort, quite the opposite, but to continue to be amazed about, and empathetic with, these experiences despite all the apparent attention focused on customer service and customer experience that I see and hear about in my customer experience day job. The columns always feature a tasty smorgasbord of stunning incompetence and total disdain for customers as demonstrated by various, usually large companies, who are considered by many to be the backbone of daily life, such as banks, insurance companies, energy providers, transport operators, and telecoms, and who fall short in delivering meaningful customer service. Despite their claims of putting customers at the heart of their operations, they often only add to the many customer service roadblocks we all face on a daily basis. The familiar frustrations include unanswered emails, delayed responses, extended wait times on help lines, and the ubiquitous automated messages urging customers to seek answers on the website. In response, many firms have turned to chat agents and chatbots to handle inquiries, but these solutions frequently exacerbate customer anger and frustration by failing to provide timely, complete, or accurate answers.
Common Criticisms of Chatbots
Chatbots are frequently criticised for their lack of effectiveness. For instance, Holly Mead, writing in The Times about a negative experience with her energy provider, concluded, “All this talk of technology replacing humans in their jobs is all well and good, but when it comes to customer service, the bots just aren’t up to scratch.” Of course, poor customer service is not exclusive to chatbots; human agents are equally capable of delivering disappointing experiences. Nonetheless, scepticism persists, often fuelled by encounters in which customers struggle to communicate their needs to chatbots, particularly when prompted to “Tell me in a few words why you are calling today,” only to find that none of the suggested examples match their actual problem. These issues typically stem from a lack of customer experience analysis, poor conversational design, inadequate training, and inappropriate entry questions.
Empathy: The Missing Ingredient
Human agents often begin interactions by asking, “How can I help you?” allowing customers to freely describe their situation and feel understood. Yet, even with human contact, there is no guarantee of genuine empathy or understanding. A global research study by Genesys indicates that contact centres struggle to offer empathetic service, with fewer than 10% of agents worldwide considering empathy, quality, and listening as their strongest attributes. Empathy, in customer service, is the ability of an organisation to resolve issues by truly understanding the customer’s perspective and acting decisively and promptly in the best interest of the customer. According to a 2022 Dixa study, an overwhelming 96% of customers consider empathy from customer service agents to be “important” during support interactions. This gap between expectation and reality raises the question: Can chatbots be trained to deliver empathy as effectively as human agents, and should organisations even attempt this?
The droids won’t take over the earth, but they’ll move into more neighbourhoods
Industry projections suggest the adoption of automation is inevitable. Gartner predicts that by 2026, 20% of agent interactions will be automated, up from approximately 1.6% today. Daniel Connell, VP Analyst at Gartner, notes that automating entire interactions can yield significant cost savings, but even partial automation—such as using AI to capture a customer’s name, policy number, and reason for calling—can reduce human agent involvement by up to a third. This approach is especially prevalent in insurance and financial services, where companies deploy conversational AI through Intelligent Voice Agents (IVAs) to handle these tasks more effectively. Unlike traditional chatbots, which rely on rigid, rule-based conversations, IVAs understand caller intent much more effectively and enable free-flowing dialogue in natural language, encouraging greater customer engagement.
The Importance of Conversational Design and IVA Training
Successful deployment of IVAs hinges on thoughtful conversational design and thorough training. While IVAs may never fully replicate the emotional depth of humans, they can be trained using call recordings and agent feedback to recognise utterances, intentions, entities, and customer tone—all hallmarks of human conversation. This enables IVAs to respond appropriately and empathetically. For instance, when a customer contacts a bank or insurer in distress, such as reporting the death of a loved one, a well-trained IVA can detect emotion in the caller’s tone and respond with sensitivity, acknowledging the loss and gathering necessary information before escalating to a bereavement counsellor or scheduling a follow-up call.
Developing IVAs for Natural Interaction
IVAs are not a universal solution, nor can they replace the human touch in every scenario. However, they prove valuable when resources are stretched, call volumes increase, and 24/7 service is needed. Implementation requires careful planning, design, and testing. Key steps include:
- Clearly defining the use case and ensuring the objective is improved customer and colleague experience, not merely cost reduction.
- Designing interactions to mirror human-to-human conversations, aiming to help customers forget they are not speaking to a person.
- Utilising existing call data and agent input to build a robust library of utterances, intents, and entities for more natural dialogue.
- Facilitating easy escalation to human agents or providing call-back options when necessary, especially after hours.
- Testing extensively, ideally internally before launching a pilot.
- Starting with cautious, simple use cases, such as after-hours calls, and gradually expanding deployment.
- Ensuring a reliable handover to live agents, complete with context and interaction history.
- Monitoring, tracking, and reporting on IVA performance and regularly reviewing for improvements as processes and desired outcomes evolve.
Conclusion: Finding the Balance – The True Measure of IVA Success
The key to successful integration of IVAs lies in a gradual, well-defined approach that balances automation with the human touch. By enhancing customer experience and delivering the empathy and understanding essential to positive interactions, organisations can ensure that both automated and human agents contribute to meaningful customer service—regardless of who is doing the talking or listening.
At the heart of any IVA deployment lies a fundamental question: has the customer achieved their goal with minimal effort, and was the process as straightforward—if not more so—than engaging with a human agent? IVAs are not intended to handle every single customer interaction; ultimately, it is the customer who determines whether their experience was positive, making them the final and only judge of success.
However, consider the alternative many customers face: enduring lengthy queues, listening to reassurances that “your call is very important to us”, only to reach the end and discover that the system provides no satisfactory answer. In these scenarios, time is wasted, and frustration grows. In contrast, an IVA can offer a prompt response, potentially resolving queries more efficiently. Therefore, it may be worth considering a conversation with an IVA—one that could not only meet but exceed expectations, and perhaps even demonstrate genuine attentiveness to customer needs.