game of rock paper scissors
PHOTO: shutterstock

If speech solutions are going to be an essential part of your company’s CX strategy, how do you choose the technology that will provide the most benefit?

"It’s useful to differentiate between skills or actions (here, VUIs) and conversational intelligent virtual assistants (IVAs),” said Erin Abler, principal conversation designer at Mobiquity. “A skill/action (VUI) can be thought of as an app built for voice-first or voice-only experiences. These usually work best as a customer-focused solution for a relatively narrow feature set, like a banking skill that lets you check account details with just a few spoken words. An IVA is an AI-driven conversational agent that’s more deeply integrated into a larger software platform or enterprise solution. It's usually paired with machine learning, which makes it smarter at extracting meaning within a specific domain.”

So, Abler explained, the choice between these solutions is largely a question of your audience needs as well as your current capabilities and scale. If your business anticipates a voice-specific use case, like hands-free communication while driving, that’s where you should start. If you’re looking more holistically at how to reach customers with conversational UIs, the suite of tools available with an IVA will likely outstrip the reach offered by a voice-first platform like Alexa. IVAs often incorporate options for human review and issue escalation, which aren’t widely available yet through VUIs.

Intelligent Virtual Assistants Offer More Uses

“The comparison isn't really about the technology, but about the user experience you can create with each,” said Jen Snell, Verint Intelligent Self-Service vice president of product marketing. “Generally speaking, IVAs are multimodal interfaces, which is helpful for accessibility. So in any instance where you might need to accommodate a person with a disability, IVAs are often going to be the best choice. In the event that the technology is being used for transactions, the IVA once again is a better choice because it can provide a visual confirmation that the intended task has been completed — such as displaying a confirmation number.”

Snell added that IVAs are generally a better option if the tasks the user is trying to accomplish are multi-step or complex in some way. For example, when booking a flight, most people want to see a variety of options before they make a choice. That's difficult to present with a single-faceted voice interface. In other words, if a real conversation is needed, the IVA is better.

“VUIs are interesting because their relative minimalism is actually their strength,” Snell added. “That's why the many smart-home uses of this technology are so compelling. For low-stakes tasks like controlling lights, playing music, or adjusting a thermostat, the action the machine takes is a definitive confirmation that the task was completed. So the primary goal of the VUI is to create a low-friction user experience, which it does beautifully. Unlike a multi-step conversation, this is a call-and-response interaction."

Related Article: Getting Started With Voice User Interface Design

VUI Excels for Simple Tasks, IVAs Go to the Next Level

While adding a VUI can provide enhanced CX, designing an IVA instead can bring voice-driven CX to the next level, according to Donna Fluss, founder and President of DMG Consulting LLC, a provider of contact center and analytics research firm.

VUIs are limited to structured queries and responses, while IVAs use natural language processing (NLP) and natural language understanding (NLU) so a customer can use different types of terminology that mean the same thing rather than using more structured queries.

When implementing an IVA, Fluss recommended:

  • Make it as clear as possible to your customers that they are interacting with an IVA or bot, not a live agent.
  • Make the IVA experience as straightforward and effortless as you can.
  • Identify the top 20% of self-service activities that typically address 80% of customer issues/inquiries and make them part of your IVA.
  • Keep the conversational aspects of the IVA as natural and friendly as possible, in all channels.
  • Leverage customer data from all available sources to personalize IVA interactions and have your IVA “remember” customer preferences, and build the integrations that will allow your IVA to provide such experiences.
  • Make sure that it knows your customers’ preferences and allows them to interact in their channel of choice.
  • Optimize the IVA for each channel it supports, but provide the same information and options to promote a consistent customer and brand experience.
  • When designing the IVA conversation, look at interactions from your customers’ perspective, prioritizing their needs and using their language.
  • When the IVA does not understand what it is being asked to do, script it to apologize and transfer the call to a live agent. The IVA should always explicitly notify customers when they are being transferred.
  • Customers should be permitted to request a live agent at any point in a conversation. Additionally, build in transfer logic so customers are transferred automatically when the IVA decides it’s necessary (and as noted above, tell them).
  • Establishing a systematic method for updating and maintaining the knowledge used by the system is essential, as a fresh, current IVA is one that customers will continue to see value in interacting with.

Related Article: Intelligent Virtual Assistants: Improving, But Still Not AI