Conversational banking has become the term du jour when banks and other financial services providers discuss Artificial Intelligence (AI). Major global banks have created proprietary chatbots for mobile apps, while other banks have chosen to work with vendors to install bots on their websites. Unfortunately, not every banking bot offers the same sophisticated level of conversational prowess, and these bots are quite different from more advanced cognitive AI platforms. In fact, the vast majority of systems installed today offer very little cognitive intelligence, almost no automation and do little more than provide answers to frequently asked questions.
In this post we’ll examine the most common missteps banks have made implementing conversational systems. We’ll discuss what was done incorrectly, what should have been done and why it matters. If your bank is interested in deploying an AI system, working with experts in the conversational AI space can help you avoid these pitfalls and ensure a proper final product.
No. 1: Choosing Solutions that Benefit Primarily Shareholders, Not Customers
If you’re considering an AI implementation and the end result doesn’t help customers accomplish their goals, the project should be re-examined. Every bank wants to save money by automating basic customer-employee interactions, and that in and of itself is a worthy goal. For example, no executive wants their contact center fielding thousands of calls on ATM locations. Conversely, if your main goal is to use AI solely to automate FAQs, you could ultimately end up with an unsophisticated product that no one wants to use.
Here’s why: Your customers are not aware, or frankly don’t care, about your banking bot’s limitations. If they receive an answer to their initial question, they’ll likely ask a more complicated follow-up or ask if they can make a transaction. A basic chatbot will field those follow-up questions in the only way it knows how – by escalating them to human operators. In the end, users have poor experiences, human intervention is still required (which was one goal of using a bot in the first place) and customers are likely to turn back to human contact methods in the future, rather than rely on an ineffective bot. In essence, you’ve invested in a project that drives customers toward the experience you were trying to avoid.
No. 2: Choosing Simple Tools for Complex Tasks
You should investigate more advanced AI solutions – particularly ones that are cognitive-based — that are skilled enough to execute tasks based on expert, data-based decision-making. To get started, you should identify processes that are high volume and apply to common business issues. In other words, you should target the most frequently occurring or repeated customer issues, ones for which an advanced AI solution can deliver results. That’s because AI systems are most valuable, particularly in the short-term, at helping businesses improve customer inquiry response rates, handle times and first-touch resolutions, as well as locating the appropriate human worker to complete a process that can’t be resolved with automation.
For example: Customers who ask questions such as, “Should I apply for a small business loan?” or “Should I cash out my 401(k)?” are not looking for a generic scripted answer. With a cognitive system in place, your bank can leverage machine learning, dialogue variance and historical memory to provide informed opinions about customers’ questions and concerns. The system can research consumers’ banking history, access market data, perform calculations and, most importantly, inquire about their financial goals in order to render educated recommendations.
No. 3: Rushing to Deployment
Practice makes perfect, even for digital workers. Cast a wary eye toward any vendor that tells you that their AI system can plug into your existing IT ecosystem and be customer-ready within hours. Installing a cognitive AI banking solution and training it to achieve your end goals are vastly different, albeit connected, scenarios.
With continuing advances in cognitive AI solutions such as Amelia, you can find solutions that follow strict banking processes, have an expert understanding of banking terminology and offer APIs that integrate neatly with your other systems. However, you still need to test — and test and test — each of these processes and actions in order to ensure that the system doesn’t fail when you most need it to succeed. As with any human banking expert, an AI system requires levels of brand-specific orientation, training and mastery in order to generate value.
No enterprise-level technology implementation is seamless and without its starts and stops. However, by avoiding these basic implementation missteps, you will set up your bank for long-term success, customer satisfaction and optimal return on your AI investment.