Banking is ready to step up the conversation. With tap-and-swipe financial services already firmly established, voice user interface (UI) stands poised as the next mainstream banking channel.
For all of its futuristic appeal, voice UI is a natural medium for banking, far simpler to use than typing or even swiping. Because banking customers regularly conduct their business while on the move, voice UI accommodates their need to stay mobile and juggle multiple tasks.
AI makes voice smarter and more natural
Voice isn’t new. Think how long you’ve been hearing the instructions on interactive voice response (IVR) systems (“Press or say 1”). But artificial intelligence (AI) makes voice interactions feel smarter and more natural. Conversational AI uses natural language processing to decode our slang, synonyms, and accents. Machine learning ensures that voice systems continually improve.
Thanks to the rapid advances of AI-driven voice, conversational AI is fast becoming a fact of life. Twenty-percent of U.S. adults have access to smart speakers in their homes. AI is even making voice warm and fuzzy: Google Assistant’s new “polite mode” reinforces use of “please” and “thank you.”
What voice can do for banks
For banks, AI-enabled voice channels can now answer a wide range of customers’ questions and perform tasks for added convenience:
What’s the industry’s best CD rate right now?
Where’s the branch nearest to me?
Can you make me an appointment with an advisor?
Do you have any offers for me with partners such as insurers or fitness centers?
Flexibility is one of voice’s strong suits. In addition to its growing popularity with consumers for purchases and information, voice is a natural platform for introducing personalized products. Once your organization’s voice capabilities are ready, you can embed them into other environments, so customers can, say, pay their cable bills while watching TV, or their auto insurance premiums while driving. Points of engagement go up — and costs often go down. One banking client we worked with shaved 20% off its contact center costs by implementing a better voice channel.
Ready, set, speak
Implementing voice UI isn’t technologically complex. A cloud-based bank can launch an Alexa app in as little as a month.
Voice is becoming an essential part of the customer experience (CX) foundation that smoothly guides banking customers in and out of desktop, mobile and contact-center channels. The CX foundation consists of an orchestration layer supported by application programming interfaces (APIs) and analytics. For banking customers, the CX foundation is critical for making a seamless experience. Using the mobile app and want to speak with the call center? The orchestration layer transfers you to an agent, who knows who you are and the channel you’ve come from.
A roadmap for getting started with voice UI
Like any new channel, voice UI needs careful building from the ground up so it works holistically with all of your bank’s channels. The best voice UIs succeed not as simple add-ons to other channels but because they reimagine the voice-first experience.
Here are steps that your bank can begin to take today to prepare for voice UI and ensure its success with customers:
Get to know the people who bank with you.
The most successful voice UIs start with the basics: They know who they’re speaking with. How well acquainted is your bank with its customers, demographically and personally? Perhaps more important, how much of what you know is based on data versus assumptions? To make the most of the advantages that an AI-driven voice channel offers, take the time to conduct an ethnographic study. Know who your customers are — not in the sense of KYC and compliance, but as people. Learn their likes, dislikes and preferences. With solid data in hand, you can begin planning your voice options based on segments. For example, you might want to segment out younger customers to whom you can push more offers. Or create options for 75-and-older customers, who find voice-based UI to be a welcome alternative to typing and touchscreens.
Understand your customers’ intentions.
Natural language processing converts words into intentions, which are the responses and questions that your customers will exchange with the voice system. Intents are an incredibly important part of voice UI. Our banking clients often assemble libraries of as many as 4,000 intents. The intents can range from simple (“Ask me later”) to more complex (“When can I cash out my CD?”). Purchasing libraries of pre-configured intents can save you time, but we recommend keeping the process in house. The reason? Intents are windows into your business. What questions are being asked most often? Which customers are asking them? Although documenting and understanding intents can take six to nine months, the upside is the opportunity to not only gain a thorough understanding of your customers’ interests but also to control the conversation — literally. Using AI tools enables your organization to identify intents from pre-recorded conversations and also to hone in on the most common and relevant requests so that your intents grow richer and more useful to customers over time. The more you know about your customers and their intents, the better engaged you can design your voice UI to be.
Create a hierarchy of responses.
Straightforward intents such as balance inquiries are simple for voice UI to handle. It takes a carefully designed hierarchy of responses to handle inquiries and requests that require complex processing to resolve. Hierarchies can span multiple interactions as they lead customers through the resolution of an issue. By analyzing past interactions, you can identify the sets of activities needed to resolve an issue. For example, customers calling about lost or stolen credit cards take a series of steps that asks them to notify the bank, opt to reorder cards, and select delivery options. Resolving the lost card issue generates multiple requests across departments and systems. To ensure value for customers, it’s critical to take time to understand the hierarchy of potential responses and the APIs that will support a seamless experience.
Implement a feedback loop.
Implement a feedback loop so what’s learned goes back into the system. AI needs feedback loops so the system can get smarter over time. The loops help you identify customer questions and comments that your voice UI can’t yet answer. For example, the word “yes” can be said in many ways: “OK,” “Fine,” “Yeah.” The AI-based feedback loop processes those learnings back into the intents.
Review your bank’s legacy infrastructure.
Voice has no concept of delay, so your infrastructure needs to be prepared to respond at conversational speed. Unfortunately, the need for speed often throws legacy architecture limitations into stark relief. Is your system able to offer real-time responses? Can it scale? We counsel our clients that an engaging, top-tier UX is likely to expose a sagging, haphazard back-office system. If a customer tells a bot they want to stop a specific check number, and the system says “Thanks, we’ll get back to you,” that’s typically a sign that a manual intervention is required — not a good look for voice UI. When your organization is ready to implement voice, don’t neglect a careful review of process and infrastructure.
AI-driven conversation is a natural addition to digital banking channels, and when done right can be a clear differentiator in a quickly shifting marketplace.
This article was written by Anil Lakhanpal, Senior Digital Banking Partner, Cognizant Consulting.