Why Chatbots Will Replace IVR in Banking

March 6, 2017

Everyone knows the frustration of talking to banking customer service on the phone. You call in with a straightforward question and get the dreaded interactive voice response (IVR) recording. Before you know it, you’re listening to a robotic voice ask how it can direct your call and having to give short, over-enunciated responses trying to get the system to understand what you want.

The fact is, IVR isn’t great for anyone. It’s not great for customers because it’s confusing and frustrating to use. And it’s not great for banks because it doesn’t accomplish its primary goal: reducing and shortening customer service calls. How often have you gotten frustrated with an IVR and pressed “0” until you were connected with a human agent?

Artificially intelligent chatbots are one possible solution to this decades-old problem. They have the potential to provide a smoother, more intuitive customer service call experience.

But not everyone is convinced. Ron Shevlin recently wrote a piece expressing his concern that the chatbot experience may end up being just as bad as the IVR experience. While Ron raises some good points about the issue, he’s missing two major advantages of chatbots over IVR.

Here’s why I believe chatbots will live up to the hype and deliver a better customer experience than IVR can.

Chatbots understand what you’re asking.

One of the most frustrating aspects of interacting with an IVR system is trying to get it to understand what you want. Most of the time, the system doesn’t let you simply state the reason for your call. Instead, it rattles off a list of pre-recorded menu options for you to choose from. Or if the system does allow you to say why you’re calling, it only understands a limited number of input phrases. Too often you hear, “I’m sorry, I don’t understand.”

Chatbots are much better at understanding what you’re asking for. Thanks to recent strides in natural language processing (NLP), chatbots are becoming incredibly adept at understanding user requests made in normal, conversational language. So instead of listening to a list of pre-recorded options about topics you might be calling about, you can simply say why you’re calling, and the chatbot will direct you to the appropriate place.

Chatbots also understand the subtlety and nuance of language far better than IVR systems do. A chatbot can understand the various phrases you might use to communicate your intent and immediately identify what you mean.

For example, say you call your bank to pay your most recent credit card bill. In real life, there are several ways to say this. You could say, “I want to pay my most recent credit card bill.” But what if you said “current” instead of “most recent”?

An IVR system might not understand that these two words mean the same thing. But a chatbot would.

Chatbots let you change your mind.

The main purpose of any customer service line — automated or otherwise — is to determine what a caller is trying to accomplish (their intent) in order to assist them.

One of the biggest problems with current IVR systems is that they typically use static, rule-based systems (e.g., decision trees) to determine customer intents. The system asks a series of questions in a pre-defined flow that doesn’t permit you to deviate from the script or change your mind.

The problem with this linear dialogue flow is that it doesn’t remotely resemble how actual conversation works. It’s rigid and robotic and doesn’t maintain context from one moment to the next.

When you talk to a real person, they understand the context of the discussion at all times. The conversation may change direction or topics, but the other person will still remember previous points of the conversation and factor that information into their responses.

Chatbots communicate with users in a way that simulates real life conversation — through flexible, non-linear dialogue rather than static, linear decision trees.

For example, say you want to send $20 to a friend. The system needs to understand several parameters to be able to make sense of and fulfill this request:

  • How much money?
  • From which account?
  • To whom?
  • When to send it?

Here’s how this request would look in a linear dialogue system vs. a non-linear one:

Chatbots vs. IVR
Chatbots vs. IVR

 

Now, what happens if you change your mind at some point during the conversation?

Chatbots vs. IVR
Chatbots vs. IVR

 

A chatbot can understand that $20 is a parameter for “how much” and fulfill a request to change the amount without having to start all over. This is one of the most powerful advantages chatbots have over current IVR systems. By allowing users to speak normally and change their minds, chatbots provide a more friction-free interface between customers and their banks.

Chatbots provide a more friction-free interface between customers and their banks. Click To Tweet

Best of all, because they’re powered by machine learning, chatbots are also capable of learning over time to improve future interactions. This means that even if a chatbot encounters an unexpected command, it can learn how to handle such requests in the future on its own (based on past experience) without the need for developers to write any additional code.

Conclusion

IVR is frustrating to use because it’s a rigid, inflexible system that forces users to restrict or change the way they speak. Chatbots are more effective because they’re a more flexible, adaptable system that understands natural language and lets users communicate normally, even if they use an unconventional phrase or change their mind.

Chatbots provide an all-around better experience for both customers and banks. For customers, chatbots are a faster, more hassle-free way to accomplish necessary banking tasks like getting account updates. For banks, chatbots reduce the number and length of customer service calls, leading to reduced customer care costs.

To me, it’s not a matter of if chatbots will replace IVR in banking, but when. The technology is ready and available now, and leading institutions are already investing heavily to reap the benefits.

As Ron points out, when a new technology emerges, every bank eventually catches up. But I disagree that offering a new technology offers no real competitive advantage. Early adopters of inexpensive, time-saving technologies like chatbots will likely see quick and demonstrable ROI by getting ahead of the adoption curve.



  • Simon Gornick

    Hey Keith, I agree that IVR has got to go, but while they’re all essentially using it, none has a competitive advantage over the next. It takes large banks a long time to do anything.

    Bot are one-up on IVR, no question, but the really enemy for most people is automation in general. They just want to speak to a real human. So the issue might be re-framed as “how to get people to be happy talking to a bot and not want to bang the zero key until it breaks”.

    Most IVR tasks that don’t require a human are far better handled through your mobile app > or a while labeled abe.ai bot on the web. But to make a real difference the chatbot has to cut into the number of inquiries that go human – and they tend to be the toughest.

    In my opinion, that’s where the focus should lie. Imagine a customer asking about the fees for a three day international wire and the maximum s/he can send. If we can give this kind of answer without triggering “robot rage” we’re onto something. At that point, people will start saying “that bot was great. it’s my new best friend! I didn’t have to talk to a tedious human.”

  • rshevlin

    Keith: You failed to mention another thing I overlooked in the IVR vs. chatbot comparison: The ability to gather and use data associated with the interactions. And I have written in the past that I believe the opportunity to create a sustainable advantage from chatbots comes from the ability to capture and use data. Great job with the article.

  • Pingback: How 10 Big Banks Are Using Chatbots to Boost Their Business | Abe()