Technological advancements over the last few decades have prompted the exponential growth of customer service across industries. The Amazons and Alibabas of the world have set new standards for customer expectations. With real-time updates, ease of information exchange, and round-the-clock customer service, industries are building solutions for seamless transactions. To meet these demands, artificial intelligence (AI), machine learning (ML) and chatbots are becoming a priority for businesses, including the banking and financial sector. Chatbots are helping organizations reduce costs, provide quick services, offer transactional support, and serve as a medium for upselling products.
The digital-era calls for instant communication and 24/7 connectivity. Chatbots enable a better communication channel for users to engage with business services. By using application programming interfaces (APIs), chatbots integrate with data management platforms to analyze data for end users. In the financial sector, consumers of data insights range from end customers to banking CXOs.
Essentially, chatbots are artificial intelligence (AI) powered virtual assistants that are trained with a conversational dataset to respond to questions. Initially, chatbots were trained on a fixed set of questions (rules-based chatbot) with which customers could interact like a frequently asked questions (FAQs) system. But as technology evolves, chatbots have evolved from sounding like a robot to emulating human speech. Natural language Processing (NLP) plays a major role in developing conversational analytics platform to provide real-time insights for decision making. NLP provides an efficient means of monitoring end user sentiments.
In everyday banking, chatbots find their use across varied functions and operations; eliminating the need for customer service agents. Gartner predicts that by 2020, 85 percent of customer service interactions will be handled by chatbots. The technology is being used to provide customers with a 24/7 support system, help with customer onboarding, KYC forms, resolving queries, and selling new products or services. According to a study by Juniper Research, as automated customer service evolves, by 2023 bank operational cost savings via chatbots will reach $7.3 billion. Future banks would be saving 862 million hours, equivalent to approximately half a million working years, as per Juniper.
Ally Bank was one of the first few banks to implement a chatbot for their customers. Introduced in 2015, Ally Assist – a virtual assistant within the Ally Mobile Banking app, helped customers request account summaries or transaction history, make payments, peer-to-peer transfers, deposits, and monitor savings. Using machine learning, Ally Assist analyzes spending patterns to predict customer needs and provide relevant help topics and messages. The assistant uses NLP to address common customer service queries.
Bank of America introduced Erica – a virtual financial digital assistant to allow customers to access information on account balance, credit reports, send notifications of account changes, and schedule transfers among many other functionalities. Six million people are reported to be using Erica and over 35 million client requests have been processed through their mobile app.
While chatbot technology continues to advance, it comes with its own set of limitations and learning that are changing over time. As more and more organizations develop use cases and expand functionalities, chatbots get smarter. The major limitation of a chatbot is still in its dialogue capabilities. Conversations with a bot do not feel natural, is impersonal, and lacks empathy and context. Bots cannot process multiple questions or conversations at the same time. They are programmed to give the right answer only when they are asked exact questions. In the real world, this creates a problem wherein different dialects, accents, and jargons/slang aren’t picked up by chatbots.
To better understand these complexities, chatbots are integrated with natural language understanding (NLU). A subset of NLP, NLU focuses on handing a narrower data set, converting unstructured inputs (human speech) to a structured form that is machine-readable (structured ontology). NLU communicate with regular end users to understand their intent; going beyond understanding words and interpreting meaning. With NLU, chatbots can enable better conversational flow using clarification techniques, detecting sentiments, and predicting customer questions. The bots now have the ability to understand context and deliver ‘contextual insights’ based on conversational banking. Fidor’s chatbot uses conversational AI and NLU to deliver real-time personalized and efficient interactions with users. It is one of the first digital banking service providers to implement virtual assistants within its technology stack. The chatbot is programmed to detect sophisticated language nuances to understand user requests and have a natural conversation.
As chatbots continue to advance, banks will need to shift their focus from mobile banking to a friendlier conversational user interface (CUI). Conversational interaction is seen as the natural evolution of mobile baking. For banks to succeed with this technology, a fundamental change is needed from both, consumers and banks, in the way they interact with each other. Not all banks are ready to make the shift to chatbots and consumers are just warming up to the idea of speaking finance to a bot.