Google Assistant available on 1 billion devices, Amazon Alexa on 100 million devices
Give Your Apps a Voice TM
How to Drive ROI with Voice Experiences
We’re all hearing a lot of buzz about voice experiences, also known as conversational user interfaces (CUIs). But will investing in voice deliver real benefits for your business? Here’s what you need to know to decide.
Throughout history, we’ve seen one rule proven again and again: when we find an easier way of getting something done, the old way of doing it fades into the background. That’s why voice capabilities will soon become integrated into every facet of human-facing technology.
Does that mean voice will replace screens? Certainly not. What we’re seeing instead is the emergence of VoiceCases—those moments when the easiest way of getting something done is via voice.
65 percent of people who own an Amazon Echo or Google Home can’t imagine going back to the days before they had a smart speaker. (source)
We’re looking at a tectonic shift in experience design, and we’ll see a huge and growing gap between the brands that “get it” and the brands that still insist on screen-based interactions where they no longer make sense.
How do we know this is more than a fad?
But behind those trends are a few simple facts about the way we humans work:
Human typing speed
Human speaking speed
Human reading speed
The most efficient way for us to interact with technology is often to speak our request and read the app’s response.
It’s faster to speak when you have a request (“Alexa, show me movie times today”), and faster and easier to read the results (a table of movie times).
At WillowTree, we call these mixed voice-and-visual interactions multimodal experiences.
Moving forward, the best conversational UIs will allow for both voice and screen-based input and output. That is, they’ll be multimodal.
What are the implications for business?
Until recently, voice wasn’t reliable enough of a way to communicate with computers. But that’s changed. Microsoft’s voice transcription last year achieved a remarkably low error rate of 5.1%—making it as accurate as professional human transcribers.
Increased reliability means voice interactions are quickly becoming the easiest way to get simple tasks done. And users’ expectations—and their brand loyalty—will shift accordingly.
Rethink your UX
You need to rethink all of your customer’s interactions with your products and services now that there is an additional tool—voice—at your disposal.
When the steam engine was invented, you rethink your entire approach to transportation—rather than hooking a steam engine to a carriage.
Similarly, you can’t just create apps for your company on Google Home and Alexa and think you’ve checked the “voice” box. You have to think about how and where voice will be the most efficient, most delightful way for your users to complete a task—even if what they’re doing via voice is just one part of a flow that happens across multiple channels.
Opportunities inside your company
If employees at your company spend any time inputting information—completing forms, searching for parts, providing customer service, searching for customer records—voice technology is going to make your team more productive.
1 hour of paperwork only takes 20 minutes via voice, saving—
Each year, per employee
Multiply that by the number of employees at your company, and it’s easy to see how voice can enhance productivity—while removing some of the stress of performing repetitive tasks.
Work-related tasks that you can enhance with voice:
- Querying large databases
- Composing emails
- Locating a part while performing a repair in the field
Opportunities with your customers
In order to discover the most valuable VoiceCases for your customers, you’ll need to do a deep dive into your customers’ specific wants and needs. But no matter what industry you’re in, the trend is clear.
Voice is going to have a profound impact.
49% would prefer voice assistants over human interactions in shops/call centers (source)
20% say they’d rather use a voice assistant instead of visiting a shop or bank branch (source)
One quarter of voice assistant users are 55 and older (source)
In April 2017, Bing Ads claimed that the number of Britons using voice search on their mobiles to book hotels rose by 343% from the previous year (source)
30% of smart speaker owners say the device is replacing time previously spent with TV (source)
32% of IT pros surveyed expect intelligent assistants to automate help desk tickets, improving customer service (source)
Across industries, customers will be able to make requests three times faster. That means transactions lose friction, conversions improve, and customers are happier.
It’s a simple fact: customers will return to an easy shopping path over a difficult one.
The analytics goldmine
Voice assistants not only streamline communications with your customers and employees, but let you learn from those interactions virtually in real time.
If someone leaves your website without buying anything, you can often only speculate as to what went wrong (or try to learn after the fact, via a survey). With voice, they’ve literally told you what they’re looking for, even if your voice app wasn’t able to give it to them.
Alexa, for example, can be a very low-cost way to learn how your business can deploy voice successfully, because it provides relatively robust analytics. Read our CEO’s thoughts about using Alexa skills to gain insights from your customers.
How to Get Started
1. Know the current state
As with any design or technology initiative, you need to start with the user and the current state of their experience. What technology is available to create voice experiences? What are your customers’ expectations of those experiences, and how can you meet or exceed those expectations with a digital solution?
Voice technology has been evolving rapidly, with new platforms, new devices, new industries, and new integrations cropping up regularly.
Take a look at some of the key things that have happend with voice in 2019.
Recent changes in the voice landscape
Apple launches AirPods with voice-activated Siri
Bank of America announces 6 million Erica users and 35 million requests
Amazon rolls out HIPAA compliant Alexa service with 6 new healthcare skills
Google Assistant arrives in Sonos One and Beam in the U.S.
Amazon introduces Alexa Conversations and Cross-Skill Goal Completion
BBC announces upcoming launch of new custom voice assistant
McDonald’s acquires Apprente, voice-based conversational technology company
Google launches Pixel 4 phone with Google Assistant built in, with all processing on the local device
2. Find use cases
The unignorable VoiceCases
WillowTree conducted a survey of 824 people (all across the U.S.) to get a pulse on which voice use cases were resonating with users. We asked people to rate fifteen use cases on two criteria: usefulness and efficiency. Here’s what we found.
Based on this research and on what we’re seeing in the market, voice is poised to become the preferred interaction model for at least three use cases that currently happen primarily via screens.
|Specific Search||Searching for a known item or piece of information||Pulling up media, directions, weather, FAQs, finding a part in inventory|
|Composition & Logging||Not suprising, but newly feasible; The impact will be huge||Writing emails, completing forms, composing lists|
|Coaching & Instruction||Anything that requires guidance, especially when there is a benefit to being hands free||Learning a new skill, getting guidance in the completion of a task|
Offer a Goldilocks feature set
3. Build the foundation
Regardless of the use cases you’re focusing on, or even the platforms you intend to start with, there a few things you should do to make you positioned to move quickly as this space continues to develop.
Take a platform-agnostic approach
The form factor, capabilities, and market share of voice-enabled devices will continue to shift rapidly over the coming years. Design a backend for your voice ecosystem that can support voice applications wherever they are: whether that’s on Google Home, a car, or a device that has yet to be invented. You should also have APIs that enable easy access for voice applications.
Start documenting knowledge and capabilities
That platform-agnostic backend needs to be able to fulfill on the use cases you’ve identified. In order to do that, it needs two things: knowledge, and capabilities.
- What can my bot do? (e.g. Help a customer find the product that is right for them)
- What does my bot need to know? (e.g. product prices, product attributes, product availability)
Gather training data
You need content to “train” your AI what your users mean when they use specific terms. This can come from any number of places; some of the most common places to look are in-person user interviews, customer support transcripts, internal or customer-facing knowledge bases and blog content.
In-person user interviews can yield lots of valuable data you can use to train your AI.
Reuse existing content
By the same token, you should audit all that customer-facing content for useful solutions to users’ problems. If you’ve surfaced use cases in your discovery process that you don’t yet have a solution for, you’ll need to develop new content to address them.
4. Design and build your simple application
So you’ve got some use cases in mind and you’ve got a backend that can support the conversations you need to have with your users. It’s time to launch something into the world!
Choosing the right platform or platforms to launch on is a complicated process—far too detailed for this discussion. But there are a few key things to consider:
- What platforms is your audience currently using? (Alexa, Google Assistant, Bixby, Facebook Messenger)
- Which platforms have the right combinations of capabilities you need to make the experience great? (e.g. Do people need to be able to view products?)
And a few best practices:
- Have an established personality for your voice apps (a “voice for voice”), but don’t get too hung up on this.
- Establish KPIs in advance of building the product
- Run regular user testing to generate further training data and hone the conversation structure
- Design flows not in isolation, but across devices—this allows you to create a fully orbed “multimodal” experience