The first virtual assistants were mostly special purpose. When we deployed Sprint Voice Command in 2001, it was used by millions of users everyday, primarily used for voice dialing, and used for other tasks such as news, weather, stocks, etc. less frequently. Back then, voice technology worked great for a limited set of tasks, some of which, like hands-free voice dialing in the car, are very important.
Twenty years ago, the mobile phone was a great communication tool enabling one to be in voice contact all the time. Email, texting, calendar and other tools were early to migrate to the mobile. Mobile data was excruciatingly expensive, priced in dollars per megabyte. The very early days of mobile.
Then, in 2007, the iPhone was announced.
Everything changed. The iPhone with its touch-screen user interface was revolutionary in ease-of-use. And since then, our expectations and uses of our devices has exploded.
Voice was great for voice dialing, accessing information sources, adding an event to your calendar. But it was not ready for the greatly expanded models for what our mobile computing devices enabled — web browsing, games, shopping, etc., etc.
Finally, voice technology has caught up. Voice is ready to be a primary user interface.
Smart Speakers Are Driving Adoption
Fast forward to Apple Siri, the first of the current generation virtual assistants, released on iPhone eight years ago. Following Siri, the Google Assistant is three and a half and Amazon Alexa is 5 years old. On special purpose “smart speaker” devices—Amazon Echo, Google Home, Apple HomePod—Amazon has the largest market share, with Google second, and Apple and others far behind. But Amazon and Google have equal share of current smart speaker sales and Google projected to have a significant lead by 2020, as seen in the chart below from Voicebot.ai.
But smart speakers are not at all the full story. Smart speakers are a fast growing market and voice input is primary user input modality. But those same virtual assistants–Siri, Alexa, and Google Assistant– are on a far greater number of other devices. Thus, the total devices with voice-interactive virtual assistants is not just the 100 million smart speakers, but now around 2 billion devices, mostly mobile phones and tablets!
Virtual Assistants Are on 2 Billion Devices
Siri, in addition to being on the relatively small number of Apple HomePods, is on iPhones, iPads, the Apple watch, Apple TV, and Macs. The vast majority of Google Assistants are found on Android phones, as well as watches, smart speakers, etc. Alexa is not on a billion mobile phones and tablets like Siri and Google Assistant, but is embedded into a growing number of smart home and other devices, and on a few phones and tablets from Sony, Motorola, Acer, LG, Huawei and others.
All told, in early 2019, Google Assistant reached over 1 billion devices, with Siri over 500 million and Amazon lagging at 100 million. And the numbers are rising fast. Smart speakers are quickly getting users familiar with voice user interaction and, with that introduction, users will increase use of virtual assistants on all their devices.
Why Is Voice Now Moving Mainstream?
One question is “why now, after so many years of hype, is the voice moving into the mainstream?”. A key driver is the improvement in accuracy of speech recognition combined with advances in natural language understanding. In Loup Ventures’ annual testing of smart speaker virtual assistants, virtual assistants can now correctly answer questions over 90% of the time. Google Assistant leading the others in understanding all 800 questions and answering 93% correctly. Compared to the test a year previous, Google Assistant performance increased by 7%, showing that the technology continues to make strong improvements.
Will virtual assistants soon be able to answer all our questions? Well, Google search continues to improve on its ability to understand and find relevant results to your queries, but sometimes makes errors requiring us to rephrase and ask again. Virtual assistants are not as far on the learning curve, so they are getting better, but far from perfect. And #alexafail still often trends on Twitter.
Clearly, though, after many decades of R&D and appearances for point solutions, the voice user interface is fast becoming mainstream. Now on 2 billion devices, more and more of us are interacting everyday by voice for information, navigation, controlling our home, and–just the early days–shopping.
In the next article, The State of Voice Shopping, Part 2, we will explore the growth of voice shopping — what the pundits are projecting and how CMOs view and react to the opportunity.
Mike Krasner is a co-founder of Vioby, a Boston-area developer of AI-based marketing automation tools for e-commerce retailers and their agencies. Leveraging their extensive background in creating and deploying enterprise-scale voice-interactive systems, Vioby is now developing voice shopping for retailers and brands. Click here to learn more about Vioby.