What Makes a Good Chatbot
Achieving higher containment rates depends on a sophisticated virtual assistant.
Cutting-edge chatbot/virtual assistant technology can be extremely effective at improving everyday banking experiences. Yet, in 2022, less than 20% of banks and credit unions utilized chatbots. Furthermore, the research firm Phoenix Synergistics reported that only 26% of banking customers were satisfied using AI-powered chatbots. Why are these statistics so low? Steven Poling, Director of Customer Success at Clinc, Inc., blames the “legacy chatbots” that began flooding the market a decade ago across all sectors. Their rudimentary operation and inability to understand messy speech often results in a frustrating communication breakdown with the customer.
What makes a bot bad.
If a customer asks for their banking balance, legacy chatbots might direct the user to the bank’s website rather than directly providing them with their balance.
They possess no intuition or context retention, such as asking the user if they would like a balance on their checking account or savings account when a customer initially asks for an account balance. Each question posed by the user is processed as an entirely new query with no relation to the prior question, often resulting in a breakdown in communication and frustration.
They are very brittle, which means they are looking for a particular keyword. Without a keyword they cannot provide the appropriate answer.
Response time is too slow, causing the customer to abandon in frustration or try calling a live agent while waiting to interact with the bot, therefore contributing to call center congestion, which the bot is intended to relieve.
Legacy chatbots may time out if the user doesn’t respond to the bot in a designated amount of time, resulting in unanswered queries and customer frustration.
Yet banking bots are not going away.
According to a What’s Going On In Banking in 2022 study by Cornerstone Advisors, chatbots made the “Top 5 Technologies in Banking” list for the first time ever. This year, one in four financial institutions are expected to invest or deploy a chatbot.
“It has taken awhile, but the industry is coming around to the realization that chatbots—or more broadly speaking, conversational AI—have become a competitive necessity,” says Forbes Senior Contributor Ron Shevlin.
“The younger generations actually prefer interacting with a bot rather than a person, providing it’s a good bot,” remarks Poling. Although the terms are often used interchangeably, chatbots generally refer to the earlier clunky technology and “virtual assistant” to the intuitive technology that leverages NLU/NLP for conversational AI today.
This sophisticated technology is able to service customers without talking to a live person. Clinc has seen its authenticated virtual assistants for a top 5 U.S.-based bank reach a containment rate of over 85%. At that high level of containment, customer adoption and usage rate escalates quickly.
“The customers using our client’s virtual assistant are getting the answers they are looking for, such as completing a financial transaction without ever talking to a customer service representative,” explained Poling. Sometimes, the virtual assistant might answer a customer query with a deep link that navigates the user to the screen within the app where they can complete their desired action.
“We’ve recorded a 90% click-through rate when the user is presented with a link,” he added, supporting the evidence that users are following through to receive their answer rather than giving up on the virtual assistant.
Clinc’s clients continue to roll out additional features like credit score and money brief, and most recently launched the country’s first Spanish-speaking virtual banking assistant.
Big advantages for smaller banks and credit unions.
Sophisticated virtual assistants providing high levels of containment offer all banks significant cost savings. However, according to Poling, it’s the financial institutions that probably use it the least that need it the most.
Yet, according to Poling most regional banks and credit unions have no conversational AI, relying on standard channels such as an employee-staffed call center.
A successful launch and continued engagement.
With 35 branches in southern Wisconsin and northeastern Illinois, Landmark Credit Union successfully launched a virtual assistant using Clinc technology less than a year ago on its public website resulting in an overwhelming reception:
- 45,000 unique customers have used the virtual assistant at least once
- 77,000 individual queries
- Over 11,000 product inquiries that could result in revenue generation
Customers interacting with the virtual assistant most often asked for information about products, branch or ATM locations, issues with username or password, and disputing a transaction. The AI transition has proved so successful for Landmark that they’re adding it to their online banking by early next year. “Landmark was an early adopter in rolling out our Public Banking Agent (PBA) to market during the pandemic. Since then, we’ve seen continued engagement with the virtual assistant across desktop and mobile. Landmark plans to expand the availability of this technology to additional channels, including online banking and iOS/Android.”
A successful launch and continued engagement.
While the technology has continually improved since the introduction of chatbots a decade ago, there is still significant variation between current offerings on the market. An explosion of chatbot/virtual assistant providers in the last ten years provides unending choices for banks and credit unions—but not all will deliver the desired results. According to Poling, to maximize containment and customer satisfaction, a financial institution must deploy superior AI technology that meets the following criteria.
Purpose-Built for Banking – Conversational AI should be created specifically for the banking industry. To ensure timely and successful deployments, a good virtual assistant solution must have pre-built, out-of-the-box integrations to digital banking providers like those from Alkami, Finastra, Fiserv, Q2 and Jack Henry.
Function like a Live Human – An effective virtual assistant should understand messy, unrestrained speech and slang like asking “How much green do I have?” or “Can you tell me how much moolah I have?” without faltering.
Conversational Healing – If a person says, “I need 50 bucks from my account. Oh wait, make that 75 bucks. ”The virtual assistant should be able to follow that instruction and complete the task without breaking down because of the abrupt change of input from the customer.
Context Retention – If a customer asks what the balance is on his checking account and then follows that question with “How much did I spend on Amazon this month?” followed by “How many of those expenses were over $50?” The virtual assistant can retain the context of the previous questions and continually provide correct answers.
Act as a First-Line Salesperson – An advanced virtual assistant with sophisticated AI can suggest and describe products, provide product rates and fees, as well as facilitate new account openings.
Voice-First Proficiency – Today’s consumer, and particularly the younger generations, don’t want to type—they want to speak and be heard correctly.
“The ability to drill down and get specific pieces of information about their finances is extremely important to users and that’s not possible with older chatbots or even many current bots on the market today,” says Poling.
Another differentiator is Clinc’s built-in integrations with digital banking systems providers, enabling Clinc to provide user-specific data in a secure and compliant manner.
According to a study by eMarketer, 123.5 million U.S. adults will use voice assistants at least once per month this year, increasing from the previous year.
Poling cites that a sophisticated virtual assistant with a high level of containment can provide a significant ROI by diverting calls away from the call center and providing answers to the customers without ever talking to a live person.
“Every level of bank and credit union is looking very seriously into investing in conversational AI as a way to provide 24/7/365 support and deflect a portion of the traffic away from the contact centers, which are already under heavy stress and unable to meet customer demand, as well as being very costly to the financial institution,” said Poling.
“Choosing a product that understands messy human language and keeps the conversation going through conversational healing and context retention will ultimately lead to higher customer satisfaction with the virtual assistant, and even more importantly, higher containment rates to help relieve the overburdened contact center.”
Start the conversation.
Clinc’s revolutionary conversational AI has been proven successful at some of the biggest banks worldwide. The best-in-class platform utilizes natural language processing that understands how people really talk, powering exceptional customer experiences that build loyalty and generate ROI. When you’re ready to start unlocking the possibilities for your bank, Clinc is here to help.