Software gets sentimental

Modern customer-service software already has a good handle on knowing what customers want and when. Now, the tools are starting to master how they feel.

Consider the mission given to AI-powered software at Kentucky Organ Donor Affiliates, a nonprofit that coordinates organ and tissue recovery and placement.  The AI listens in on over 10,000 calls per month—from doctors, nurses, donors, families, and hospitals—that are involved in the complex and emotional process of delivering organs to transplant centers within strict time frames.  

The software transcribes those conversations as they’re happening, flagging positive and negative sentiments and analyzing other details—including keywords and voice changes—to ultimately grade the call on a scale of 1-5 for customer satisfaction.

Over the past six months, the AI tools have helped boost KODA’s customer satisfaction ratings from an average of 2.5 to nearly 4. The improvement has helped customer service representatives to better handle stressful and emotional conversations with families and hospital personnel, says William Schmidt, operations manager at KODA.

“We’re talking about death and dying, and there are emotions that go along with that,” he says. “When the system flags a call with a below-average rating, we want to review that so we can see whether the AI was picking up on emotion and grading the call lower, or whether there’s an opportunity to talk to our staff and review how they could handle a particular situation a little more sensitively.”

Software gets sentimental

Sentiment analysis has long been a table-stakes tool for businesses trying to improve customer experience. By analyzing customer opinions in aggregate, companies can gain an idea of overall satisfaction levels and identify areas for improvement. 

But conventional sentiment analysis tools have limitations. First, they primarily draw insight from reviews, feedback forms, and social media—highly biased sources of customer sentiment, as customers self-select before sharing their feelings via these channels. Second, these tools typically have limited language comprehension capabilities, hindering their ability to understand the nuance of what customers were truly saying.

Read also: Generative AI—A new catalyst in customer support

New, AI-powered tools can dramatically improve sentiment analysis, creating opportunities to gather key insights across channels that haven’t traditionally been associated with the tactic—particularly chatbots. Specifically, improvements in natural language processing (NLP) enable chatbots to comprehensively understand the tone of the customer, providing well-sorted and less-biased sentiment data. Not only that: when bots better understand customers in real-time, they provide better service, directing the conversation according to the customer’s mood and tone. 

Sentiment analysis supercharges chatbots

Consumers expect always-on service, with the number of live chat requests per agent more than doubling between 2020 and 2022. By providing instant response times at all hours, chatbots fill a key void: They might not necessarily replace human customer service wholesale, but they provide a valuable supplement when humans simply aren’t available. 

An intelligent power lies in its ability to perform complex dialectical tasks, such as recognizing the emotion—or sentiment—behind a human user’s words. In some cases, a human may have a question with a single answer where emotion doesn’t matter. Say an airline customer chats with a bot about their flight, asking, “Which gate should I head to?” The answer is simple and straightforward. But in other cases, customers may be in need of more empathetic—and potentially less clear-cut—support. This is where AI can make a difference.

Imagine that same airline customer’s flight is delayed or canceled. If the bot senses an angry tone, it may offer the customer more emotive support and even choose a remedial action, such as a discount code for a future purchase, steps to rebook the flight, or ways to receive a refund. Determining which step to take has previously been a largely human action, but thanks to a combination of AI-enabled sentiment analysis and decision-making, a bot can now make the same call with the same level of confidence. An advanced bot can even use machine learning (ML) to expand and develop its vocabulary, learning colloquialisms in real time to ensure a proper response. 

Companies are experimenting with different applications of AI-powered bots. Twibi, a digital marketing agency, recommends its clients use chatbots to improve customer experience on their websites. “AI allows chatbots to learn how to have more natural and engaging conversations with users,” says Founder Brenton Thomas. This improvement in conversation quality helps retain prospective customers and satisfy new ones. “AI-powered chatbots can revolutionize the way we interact with computers.”

Meanwhile, at VEM Tooling, an international plastic manufacturing solutions company, AI-enabled chatbots answer common questions and assist prospective customers in the initial stages of sales. “Using AI-powered chatbots has resulted in improved customer satisfaction,” said Melissa Terry, digital media manager at VEM. “For instance, a client from a distant time zone had an urgent question about how we manufacture plastic. The chatbot was able to give them thorough information and promptly allay their worries.” Terry says that this swift, accurate response strengthened the business relationship. 

As AI-powered chatbots gain recognition and capabilities, the possibilities are vast. For example, chatbots can enable businesses to unlock previously untapped revenue streams, turning service interactions into selling opportunities. Whereas past iterations of bots weren’t designed to identify a window to upsell or cross-sell a product, smart bots capable of sentiment analysis can assess when a customer might be open to such an experience. 

U.K.-based online furniture marketplace Flitch has experienced this first-hand: “Chatbots have helped raise not only customer satisfaction levels, but also revenue and conversion rates,” said Founder Daniel Ufland. By providing on-call, all-hours support at the beginning of the sales funnel, NLP-fueled chatbots engage prospective customers who might otherwise leave the site. 

Chatbots can inform future strategy 

As they mature, chatbots are no longer just for improving customer experience in the moment. Thanks to AI, they can also inform future CX strategy by identifying areas for improvement. 

Chatbots receive higher, less self-selecting traffic than online reviews or tweeted complaints. As a result, AI-enabled chatbots can be used to collect more accurate, comprehensive sentiment analysis data of overall customer experience than other feedback channels like social media or post-interaction surveys. 

Research has shown that performing sentiment analysis on past chatbot interactions can allow businesses to “objectively and automatically retrieve valuable information after and during an online service encounter”—both simplifying and improving the sentiment analysis process. 

Sentiment analysis can even help developers improve the chatbots themselves. By tracking when humans grow frustrated with bots versus when they’re satisfied, businesses can better home in on the specific technical features that make or break the chatbot experience. One museum, for example, used chatbot histories to analyze museum visitors’ opinions of artworks, in turn using the data to ask better questions about visitors’ feelings and solicit more accurate input.  

AI helps lend a human touch  

91% of consumers said they would prefer a chatbot to human support if they knew the bot could accurately answer their questions in real time. AI allows chatbots to do just that, comprehending customer intent and feelings, asking for clarification instead of misinterpreting, and harnessing ML to provide better support as they go. 

Thomas put it simply: “This new tech makes chatbots more user-friendly and accessible to a wider range of people.” AI-powered chatbots don’t just provide 24/7 service—they empower companies to meet more customer needs more efficiently. 

Kristin Burnham contributed reporting

We want to hear from you! Please send us your feedback, and get informed about exciting updates from The Works. Drop us a line: theworks@freshworks.com.