The Missing Piece in AI Programs: Emotion
In the realm of big data, there is an urgent demand for systems to evolve its artificial intelligence (AI) with emotional intelligence. Understanding customer’s emotions is key. The ability for businesses to not only see, but also comprehend how people value and go through its process via AI is invaluable.
Though it’s easy to set the technology in place, AI’s potential is lost if businesses don’t understand how to use these advanced tools to discover even the most basic emotions.
Spike Jonze once said, “Emotions are messy and hard to figure out.” In his movie “Her,” an empathetic AI system tried to understand users by the tone of their voice, spotting lies or discomfort from speech patterns without seeing facial expressions.
The days of sci-fiction are not quite here, and neither is AI’s ability to elaborately understand feelings. And to be frank, few people really grasp the everyday emotions related to customer experience and life.
Yet there are ways to apply AI to identify micro emotional attributes of an interaction, however, which provide better emotional engagement and dialogue.
Gaining strength from weak signals
Your customer’s emotions are one of the most powerful factors of a brand relationship. Think about the power behind your company’s name, does it evoke eye rolls, or does it bring back memories of seamless and efficient service. According to Forrester, a consumer can ditch a shopping experience or leave your website in almost 50 milliseconds if your experience does not meet their expectations.
What it comes down to is how AI can capitalise on what our emotions are trying to tell them, doubled by how complicated it is to translate unstructured human interactions into tangible data. In most transactions, emotions are fleeting—the anger or approval you express during the said time usually fades away after a few hours. Emotional analytics can be used to capture these brief movements to understand the state of the consumer and brand relationship. And fundamentally, how they can overcome it.
Basic conversations such as, “the agent was pleasant and knew his stuff, maybe I’ll buy from them,” or “I love how they send me a message if my package is late” are excellent indicators of brands gaining trust and loyalty. AI can be used to scour these unstructured data points from call logs or surveys, for example. The process can reveal incredibly short moments of emotional insight that may never be intentionally given as feedback.
A complicated issue
Ultimately, we know it’s hard enough to describe your own emotions sometimes, so reading others can be a whole other beast to tackle. No two customers will ever feel the same way about your service and to draw emotions, we need a solid narrative.
A good narrative requires a flexible approach. According to TTEC’s head of insights, Peter Dorrington, the best way to analyse emotions is through algorithms that can uncover underlying feelings through open-ended responses. This is helped through questions that recommend “description,” i.e. “how would you describe today’s service” or “describe this product to someone.”
“Not all narratives are equally representative,” says Dorrington. “For example, customers calling into a complaints line are likely to be biased towards negative emotions. Comments made on social media are more likely to represent both ends of a distribution, and none are likely to represent the ‘silent majority’ who are not talking with you at all.”
Narratives are essential to gain any value from the daily interactions your customers have with your brand. When you move away from the responses that base satisfaction from a 1-5 scale, you create a bigger picture.
Tying it all in
Our Customer Experience Vector (CXV) is one way to better understand the complex field of emotions. This system of measurements combines the data science of ‘who, what, when, and where’ with customer patterns and trends, with behavioural science that can help understand the ‘why’ and reach better conclusions on customer feelings.
The CXV feeds off the open-ended narratives that customers input. This in turn seeks out weak and strong signals that the customers use in their dialogue and speech. The basis of this are the positive and negative emotions that we pick up from narratives. A graph is created from those fleeting emotional moments from agitation to calmness or disgust to trust. There we can find inflection points in the customer journey.
Overall, understanding emotions will never be an easy task, No two customers are ever alike, and businesses need to acknowledge the emotional footprint they leave with audiences. While AI will never develop enough empathy to understand how we feel, brands need to understand how well it can evolve to perceive our interactions.