For Brands, Past is Prologue for Alexa and Google Assistant

-with Dan Miller

It’s been almost 2½ years since Amazon surprised the world with the release of Amazon Echo and its embedded voice assistant Alexa. Enough time has passed to analyze consumer usage data and begin to draw some conclusions.

How Are People Using Alexa?

Bret Kinsella of Voicebot.ai recently drew attention to an Alexa user study published last year by Experian. Kinsella observes that most usage can be divided into two main categories: functional tools and audio consumption. Representing the functional tools category are popular skills such as “set timer,” “check time,” and “set alarm.” Audio consumption is dominated by the skills “play song” and “read news.”

It’s interesting to note that the most utilized functional tools are “native” to Alexa. In other words, they aren’t services provided by 3rd parties that require a special (and often difficult to remember) invocation phrase. An example of a typical invocation phrase for a 3rd party skill is “Alexa, ask Stock Quote to get quote for MSFT”. The clumsy aspects and unnatural feeling surrounding this type of “conversation” has been a source of frustration for companies and brands who would like see these home metabots morph into more conversational companions.

Trailing behind simple functional tools and audio consumption is the use of skills to control the smart home. While nearly 85% of Alexa-enabled device owners have used the voice assistant to set a timer and 66% have had Alexa read them the news, only about 46% have utilized voice commands to control their lights. Here again, we’re just starting to iron out the kinks involved with using voice to take control of more aspects of our daily life. Patching together disparate smart devices takes more effort than most people are willing to expend right now.

What’s Wrong With the VUI?

Despite all the hype around the potential for voice assistants and the voice user interface, users are sticking to native skills that are simple to invoke. A GUI (graphical user interface) might aid both the discovery and navigation of voice applications. Indeed, both Amazon and Google have used email “alerts” and in-app notifications to promote new skills, actions or uses for their respective voice-first bots. Yet 3rd party providers are having a tough time getting users to discover, adopt and stick with their skills.

Native text-to-speech (TTS) also seems to be an issue, as the most loved skills don’t rely on TTS. Audio consumption skills that depend on TTS tend to receive poor user reviews and low adoption rates.

Why the Race For Voice Assistant Dominance?

Yet as the reality of VUI applications sets in, the race to control the voice assistant space continues heating up. The key rivals are of course Amazon and Google. Amazon is on a steady march to integrate the Alexa Voice Service (AVS) into every possible device. AVS has been embedded into the latest Kindle Fire tablets for months and the retailer recently added Alexa to its main shopping app. We noted earlier this year how pervasive AVS was at this year’s CES, appearing in everything from cars, to home appliances, to robots.

Google continues to expand the terrain covered by its own voice assistant. Google Assistant not only lives inside the Home hardware device, but it can now be reached from some Android devices and has merged, or superseded, Google Now.

So What Does the Future Look Like?

Several things about the voice assistant space are becoming clear. Firstly, the actions of Amazon and Google indicate that both companies believe in the critical importance of winning the battle for the dominant voice assistant. Secondly, the ecosystem where voice assistants will “live” is large and more or less ubiquitous. The dominant voice assistant will most likely greet us in our homes, from our smartphones, in our TVs, our cars, our augmented reality games, our offices, and wherever else we spend time. Thirdly, users seem to gravitate to native skills. To branch out, they are going to need hand holding. The voice assistant itself may end up being the best guide to what it can do.

These data points lend credence to the conclusion that the future belongs to one or more metabots: ubiquitous smart assistants with many embedded skills whose availability, promotion and distribution is controlled directly by the metabot provider.

Already 3rd parties must navigate certification rules strictly managed by Amazon and Google in order to get their skill or action included on either platform. One can speculate that the platform owners will gradually seek to own any application that is frequently used, especially subsuming any useful functional tools skills. At the same time, the metabot providers will seek to control content that satisfies user needs for audio consumption. It smacks of the “browser wars” that plagued the World Wide Web and the OS wars that continue to require app makers to invest in multiple development and maintenance efforts.

Metabots: Utopia or Dystopia?

Is this trend positive or negative? The trend towards monopolistic metabots certainly seems to have some risks. That model makes it more difficult for small application providers to get their voice apps into the “ears” of an audience. It even creates challenges for businesses that seek to leverage a broadly adopted voice assistant to offer voice-based services to their customers and prospects.

There are technical issues, like the aforementioned issue with monotonous text-to-speech rendering that make voice-based services less than engaging. There are psychological barriers as well because the best of the metabots fail to answer questions fairly often and individuals are finding their shortcomings to be unpredictable and, therefore, unreliable. But the key issues at this point have to do with the structure of service delivery.

As we noted well over a year ago, once the metabot turns into a gatekeeper for the public’s attention, all sorts of issues arise. Whose service gets activated when someone orders flowers, or asks for a rideshare car, or needs new car insurance? Today, the metabot decides, and this structure creates a situation where it is not behaving like compassionate, personal assistant to its owner. It is just another digital commerce channel with a limited set of capabilities.

The implication for financial services companies, retailers, travel and lodging, healthcare providers and other businesses with expectations for “Bots” to blend into a customer-first engagement models is profound. It will be hard to accomplish without making the bot into an assistant that acts on behalf of its owner.  The shiny new world of voice assistants may turn out to be brighter than we imagine. Or it may not come to fruition at all. In the meantime, businesses have no choice but to continue experimenting with voice-based applications to find out what works and what doesn’t. We urge you to share your experiences on our Intelligent Assistants LinkedIn Group, and get the low-down on the latest real-world experience by attending the Intelligent Assistant Conferences.



Categories: Conversational Intelligence, Intelligent Assistants, Articles

Tags: , ,

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.