Get out of IVR jail free: Tips to ensure AI and IVR work seamlessly together

In the last four decades, interactive voice response (IVR) has come a long way. Yet, even as the routing options and solution/response vocabularies have grown, IVR can still be a regular source of pain for customers when it leads them to a frustrating dead end – with no answer to their original query and no clue how to find one. This is where we get the notion of “IVR jail.”

While IVR jail can take many forms, in just about every case the source of the issue is a fundamental misunderstanding of the customer journey. To illustrate this point, let’s take a closer look at a few of the most common examples of IVR pain points, according to one recent Vonage survey.

• “I’m forced to listen to irrelevant options.”
• “The reason I’m calling might not be specifically listed.”
• “The menus are too long.”
• “It keeps me from getting through to a live person.”

In each one of these frustratingly common customer experiences, IVR was tasked with replicating a customer-to-agent experience, but it couldn’t execute a natural, conversational interaction that successfully matched customer expectations at each step in the caller interaction. The unintended consequence of this type of IVR experience could easily be a lost customer or even multiple lost customers if they complain to their friends or post on social media – ouch.

Can AI help solve the IVR problem?

Artificial Intelligence (AI) has captured the human imagination in the last few months for its ability to create the most human-like conversations we’ve ever had with human-built systems – generating relevant, reactive responses to questions in real-time across a variety of different use cases. It’s helping copywriters, doctors, video game designers — you name it. But what about IVR?

Emerging AI capabilities hold tremendous promise to enhance traditional IVR interactions and turn them into desirable customer experiences. In fact, many of the use cases and solutions for AI in IVR already exist.

With its usefulness not in doubt, the question for contact centers becomes two-fold:

• What can AI do to enhance IVR and solve its historical shortcomings?
• How do you implement these AI features in a way that doesn’t lead to an entirely new form of IVR jail?

Let’s tackle these questions one at a time.

Part 1: Ways AI can improve IVR delivery

Before we get into the logistics of deploying AI into IVR, it’s important to understand the capabilities of AI that can be used to enhance the weaknesses of traditional IVR.

1. Speech Recognition: Speech recognition grammar specification (SRGS) standards helped traditional IVR speech engines to recognize speech, but these speech recognition abilities can be quite rigid. Only specific sentences and phrases included in that IVR’s grammar knowledge base will be recognized. In other words, traditional IVR is designed to lean heavily on a predicted range of utterances – and therefore a predicted range of “magic words” for customers to continue on their journey to find a solution. AI-enabled speech-to-text services allow customers to move much more directly through their intended path to a resolution. Massive amounts of training data behind the scenes can enable an AI-supported IVR to learn and capture a wider range of responses and utterances a customer might be using. As terms and customer language evolve, always-on language model training happening in the background can make sure an IVR is primed and ready to recognize these new terms quickly.‍
2. Speech Understanding: When it comes to responding to customers, context matters. For example, if you were calling in to order a different size shoe, because the ones you ordered do not fit, “size 12” means something entirely different if you’re talking about the size you previously ordered or the size you need. AI’s natural language processing abilities continue to increase the accuracy of natural language understanding in spoken and chat conversations, so typos in a message with a chatbot or homonyms in common speech don’t negatively affect the IVR’s ability to correctly interpret the intent of the answer and route you to the appropriate solution.‍
3. Speech Generation: Lastly, text-to-speech and generative AI capabilities are rendering IVR studio recordings obsolete. This, in turn, allows contact centers to quickly build out more flexibility that follows a more natural conversation flow.

Part 2: Considerations to prevent IVR Jail 2.0

The benefits of AI for an IVR use case are many, but it isn’t too hard to imagine a situation where the implementation of these powerful features could lead to an entirely new form of IVR jail.

Picture this: A customer calls in and is immediately prompted for an open-ended response to the question: “How can I help you today?” The customer on the other end of the line isn’t really sure how to accurately articulate their need, and after a couple of failed attempts, starts to feel that same old IVR frustration bubbling back up.

While Intelligent Virtual Assistants (IVA) can provide a more natural dialog than any IVR solution to come before, they aren’t perfect. There will still be moments when a customer goes off script, and a human agent will need to step in to fulfill a request. That means you’ll still need to have the ability to identify that customer frustration is building and have the integrations to seamlessly hand off the interaction to an agent.

In order to prevent these types of IVR jail scenarios from arising, here are four important questions to ask during the initial solution-building process:

• How are we currently routing calls through our contact center? What journeys and intents are most common?
• What makes sense for IVA to support via self-service, and what intents need to be directly tied to live agents?
• How will you manage the handoff to a live agent to avoid sending a customer “back to the beginning”? Will the IVA provide caller context as the agent takes over?
• What sort of cognitive load is our IVA customer experience putting on the user? How can you build a conversational flow that is both intuitive/straightforward and open-ended enough to properly capture customer requests?

TTEC Digital is here to help you make the most of your AI investments

Along with our partners, TTEC Digital has been pioneering AI solutions in the contact center for more than a decade. That means we’re ready to help you solve the tough questions – because we’ve solved them before. If intelligent IVR is on your list, our tech-agnostic team of CX experts can help you assess your AI readiness and map out key considerations throughout your omnichannel customer journey. The result: you realize the promise of AI’s ability to introduce speed, accuracy, and adaptability – without finding a new way to put your customers behind bars.