In a recent Opus Research webinar, “The Shift to Customized Voice Assistants: How Will You Compete?”, we joined analysts from Opus Research and RAIN agency to unpack the results of a recent global survey of 320 business leaders in 8 key industries. The conversation ranged from a discussion of recent trends and the rapid evolution of voice AI as a tool for better customer experiences, to the ways in which leading brands are approaching the implementation of voice assistants and measuring success.
In addition to the webinar, Opus Research has compiled a detailed report, “The Business Value of Customized Voice Assistants.” In the report, survey results are revealed and discussed in relation to recent voice AI industry news and trends and the state of the voice AI technology. The report focuses on proving ROI, finding real business value in voice AI, and the methods used by top brands across industries to measure that success—as well as their plans for the future.
The panel of voice AI experts who participated in the webinar, included:
During the webinar, attendees were invited to ask their questions and participate in the discussion. Although we were able to get to a few of their questions, many were left unanswered. The following is a recap of the Q&A session with answers from our panelists to all the questions from the audience.
Companies who are looking to add voice assistants to their products or services always start with a core value. If you’re a retailer, voice search is potentially really powerful, and once you’ve established that a customer has a need, then you can make suggestions either proactively or reactively.
For instance, a retailer could respond to a customer who’s looking for a jacket and suggest that there are some pants or some shoes that go with it. That’s simply pointing out helpful things in the context of what people are saying. You just have to get it right, and voice can be the breakthrough.
Q: Mike mentioned an omni-channel voice experience that will emerge and unite multiple touch points. How does a custom assistant help that evolution?
When we talk about omni-channel, we mean that you want your assistant to be where people are and if they have an Alexa speaker, you want to be there as well.
There is a lot of value in a company having a custom branded assistant that sits within their own channels—where they have more control over the user experience and where they have more control over that data and can derive user sentiment and emotion.
There are a lot of different ways to approach this, but the end goal is for companies to make sure they’re providing a unified and consistent experience to their customers and getting as much information as they can from those interactions.
Q: How can voice technology bring more “trust” to customers?
It really comes down to personalization. If I trust that the entity I’m giving my information to is going to use it to my benefit, then I’m willing to make that exchange. If that information is going to get used in ways that I don’t agree with, I lose that trust.
And that’s at the heart of the privacy discussion that we’re having now. The relationship that a brand makes with its customers is one based on trust. You trust brands to deliver on their promises, and that extends to your information. In addition, you may not trust one brand the same way you trust another.
That is at the core of what building a custom voice assistant allows companies to do. If somebody knows more about me, they know more about what I want, and I’m going to value their suggestions more. The reality is, computers can do that better than people because people forget things and the information goes away or it’s misinterpreted.
There is real potential here to create a trusted relationship. It just has to be done in a way that creates real value and it needs to be done very incrementally, I think people want it and they’ll accept it—if the rules are clear and if the brands deliver on their promises.
Q: How to integrate mobile apps with voice input smoothly without using smart speakers?
There are two paths you can take to voice-enable a mobile app. Either you can integrate your own ASR and NLU or you can work with an independent voice AI platform provider, like Houndify, to create a custom voice experience that can include a branded wake word. If you already have ASR and need NLU, technology platforms like Houndify can also provide the individual tools and technologies you need to create your own voice assistant.
Q: Is a custom assistant one that’s free from Alexa/Google platforms?
Yes, exactly. When people hear voice assistants, they immediately think of Alexa and Google. They’ve built a notable ecosystem where you can put skills and services on as part of that smart speaker.
That can be valuable if you’re a service provider—if you’re a bank or a retailer and you want to enable your capability through the Alexa speaker. The challenge becomes discoverability. How does your brand show up? What is your relationship to the customer?
It’s not like the app store where you’re using an app and the rest of the phone goes away. Instead, you’re interfacing directly with the company. If you’re voice-enabling your service through Amazon or Google, you’re really still underneath them. Since it’s a much more obscured experience, brands naturally start wanting their own assistant.
You see a company like Pandora who exists on Amazon and Google with millions of subscribers, but they’ve also voiced-enabled their own app using our technology. So if you go into the app and say, “Hey, Pandora,” you’re talking directly to Pandora.
They get to see the user data and the experience. They’re getting smarter about how they show up on voice because they’re ultimately controlling the experience.
Q: What’s your view on domain-specific vs. generalized vocabularies for ASR in voice assistants?
The narrower the range of knowledge, the more accurate your ASR can become. For instance, if your voice assistant is developed specifically for the medical environment and a doctor is dictating notes using drug names or industry jargon or shorthand, an ASR with domain specific knowledge will make assumptions based on the context of the options within a medical setting and deliver high accuracy rates.
On the other hand, generalized voice assistants will hear the same information and choose from a range of unrelated domains that make it harder to accurately recognize the meaning behind the words—leaving room for gross misinterpretation and low rates of accuracy.
Q: Sometimes, I find Google Actions or Alexa Skills are so limited in what they can show. How can these provide a consistent brand experience?
We agree that third-party platforms are limited in scope and ability to deliver a consistent brand experience. It’s really difficult to be consistent when you don’t have control over the experience or the user data.
For brands, there are three paths: build it yourself, live exclusively on Amazon and Google—but ultimately deal with the consequences, or you can partner with an organization like SoundHound or get consulting advice from agencies like RAIN. In terms of offering a consistent brand experience in every channel, companies are realizing that they can’t just live on these smart speakers. If you pick up the phone and you’re talking to an automated assistant, can that be the same assistant that is in your app or in your retail store kiosk or in the drive-through?
Custom, branded voice assistants allow companies to determine their own voice roadmap and design and improve experiences based on actual user data and feedback, and that’s a strongly emerging strategy for voice AI.
Unless you’re building a voice assistant experience for an independent speaker, your skill or action will be part of the third-party speaker ecosystem.
Q: Do you see a rising demand for custom, banded voice AI vs. human voice?
There is a rising demand for improving the customer experience through faster response times, more choices, and greater convenience. In some industries, like restaurants and hospitality, there is a growing demand for touchless, contactless transactions.
To achieve the ultimate goal of a consistent customer experience across channels, companies are seriously considering the role of a branded voice that will deliver their image, tone, and personality to customers and prospects at every turn.
Q: When will we have a personal AI voice assistant like in the movie “Her”?
We’re not there yet, but it all goes back to that data piece and why it is so important that brands are learning from customer voice experiences, including: “What did I glean about the way they’re speaking to my brand and what questions they’re asking?” so that we can then use that to flip those questions right back to them in a more personalized and proactive way.
Using the existing technology, compelling voice experiences are possible now—but only if companies are willing to customize their voice assistants to meet specific user needs and desires.
Q: Why would users choose to talk with voice agents instead of a real person or looking for the information by themselves?
Greater convenience and the ability to interact with devices or services hands-free is driving user adoption of voice assistants. Voice AI can retrieve the information faster, and more accurately, than a human can, since manual search generally brings up pages of results that need to be sorted. A voice AI system can take context and user intent into consideration even as the user is talking and narrow their search to deliver specific and even personalized results.
Q: How do you validate that the voice app increased user engagement and satisfaction?
Our survey showed that companies are comparing Customer Satisfaction Scores (CSAT) and Net Promoter Scores (NPS) from before voice experiences and after—and finding significant, measurable results.
Q: How would you validate that your voice achieved the preferred matrixes?
User testing is a key component to voice assistant design, development, implementation, and ongoing optimization. From the beginning user testing should be a key element in the workflow of voice assistant creation.
Depending on your business goals for a voice assistant, performance can be measured in many ways. Many respondents to the Opus research report stated that their organizations were measuring voice assistant success by monitoring and recording customer satisfaction (CSAT) and Net Promoter Scores (NPS).
When you set your KPIs for the custom voice assistants, you will want to make sure they’re measurable goals and determine how you’ll measure progress and at what intervals.
Q: How does the recognition and collection of sound (not just voice) figure into this brand discussion?
This survey was focused solely on the progress toward the implementation of voice assistants and the results of any implementations already in place. While sonic branding was not part of the discussion, we recognize that all sound contributes to the brand’s identity and recognizability in the market.
Q: Were automotive companies included in the survey?
Since we work very closely with the leading automotive companies, we didn’t feel the need to get the pulse of the industry. Although they weren’t included in this report, we have many resources on our website where you can get more information.
Q: I work on voice for in-car use-cases. How can I convince management that we need to rework our process so there is a voice-first approach?
We’d love to help you with that. You can share some of our success stories for brands like Hyundai, Honda, and Mercedes-Benz or share statistics from our website (listed above). Feel free to reach out to our sales team to discuss partnership opportunities.
Get the whole voice AI picture
During the webinar, the panel of voice AI experts discussed:
- The deployment of voice assistants across industries and the trends for how and why most businesses begin by voice-enabling a mobile app.
- Differences in voice assistant implementation by industry and the impact of recent global events.
- How voice assistants drive value in organizations, why convenience and speed top the list, and what that means for future iterations of the voice experience.
- The business drivers behind voice assistant implementation.
- The preferred metrics for measuring the success of voice experiences.
If you missed it live, or if you want to see it again or share with a colleague you can view the webinar in its entirety here. Ready to take a deeper dive into all the findings? Visit our dedicated landing page to get more commentary on the findings and download the full research report for free.
At SoundHound Inc., we have all the tools and expertise needed to create custom voice assistants and a consistent brand voice. Explore Houndify’s independent voice AI platform at Houndify.com and register for a free account. Want to learn more? Talk to us about how we can help bring your voice strategy to life.