Google Home 6x more likely to answer than Alexa, but it doesn’t matter

Chris Burns - Jun 27, 2017, 10:49 am CDT
2
Google Home 6x more likely to answer than Alexa, but it doesn’t matter

A study published this week found that Amazon Echo with Alexa – and Alexa in general – was not nearly as ready to answer questions as Google’s Google Home. The systems were tested by New York firm 360i who suggest they’ve used their own proprietary software to test both home assistant technologies. In their first brief on the subject, they say that their Voice Search Monitor (VSM) software shows Google Home as “six times more likely to provide an answer than Amazon Alexa.”

In each round of testing, both Google Home and Amazon’s Alexa software were asked the same 3,000 questions. That’s plenty for an initial test of two systems such as these – but what about bias towards one system or the other? Is Google Home meant to be a different sort of assistant than Alexa? Are both systems expected to answer the same questions by consumers?

SEE TOO: Amazon Echo Show Review

It seems to me that Google Home is aimed at a much wider audience than Alexa, while Alexa is aimed at Amazon users. Google Home is meant to be a device for Android smartphone users, sure, but it does not rely on the user’s need to purchase items and subscribe to services. Alexa devices do not NEED the user to subscribe to any service, but it’s far, far more useful when they’re a part of the Amazon ecosystem with services like Amazon Prime.

While the following has little to do with both systems’ ability to answer from a set of 3,000 questions, it’s also telling with regard to these device’s audiences. While Alexa is able to answer questions of a single person logged in to their Amazon account, Google Home can work with 5 individual logged-in Google users, identifying them by their voice signature. The same goes for accounts that use log-ins: Spotify, YouTube, calendars, etcetera – only one on an Alexa system, 5x users on Google Home.

The 360i test checks whether one of these software assistants can answer a given question. Instead, what should be tested is the real-world use-cases and satisfaction of the intended audiences of both devices.


Must Read Bits & Bytes