The Amazon Alexa Business
An intriguing report came out that looks at Amazon’s Alexa business and the organizations imperative to know figure out how to make money.
The report states the Alexa organization, which employs roughly 10,000 people, fell short of revenue projections north of $5.5m and only brought in $1.4m. Much of this revenue was collected off revenue generated from third-party skill and essentially tells us consumers are not yet using their Alexa’s for commerce, spending, or transactions in general.
We have been studying how consumers use their Amazon Echo’s, and Alexa in particular, so this comes as no surprise to me. In general, people are using voice assistants for basic things like playing music, setting the alarm, basic search query, etc. My thesis all along, with these devices and the voice UI, was how voice eliminated friction to do things that would require multiple steps on your phone. Saying name set alarm for 6 am is much faster and easier than using your phone to find the clock, select alarms, then choose the alarm time. A task with a voice request which would take less than 5 seconds and a task which by hand could take more than 20 seconds. Every study we have done on how consumers use voice assistants falls into the automation category.
Amazon has much more low-hanging fruit than the Alexa ecosystem when it comes to generating revenue. While Amazon has the lead in terms of the smart home share as well as the largest third-party ecosystem, unless we see significant changes in user behavior, I’m not sure how this business turns into any significant revenue for Amazon but remains an important loyalty and stickiness part of their strategy.
The Challenge of AI
We all know AI is a hot topic. And I’ve written for years that before we start talking about AI, we have to first understand and watch how machine learning as technology develops.
The desire to use AI/ML in the enterprise, specifically is high. I’ve seen several IT surveys this year and consistently the results highlight over 80% of those surveyed say they are either actively deploying some kind of AI or looking to in the next six months. But as this report highlights, it is still a difficult endeavor.
Algorithmia, which found that while machine learning maturity in the enterprise is generally increasing, the majority of companies (50%) spend between 8 and 90 days deploying a single machine learning model (with 18% taking longer than 90 days).
There are many challenges that explain this issue, plenty I don’t have time to dive into, but it is worth knowing that analytics is the most common area where companies say they have AI deployed. Most still require a team of data scientists, but this is why technologies in the works from Microsoft, Google, and Amazon to automate AI as a part of the analytics process and lessen the need for data scientists is important.
Understanding the challenges of deploying AI is also a factor in understanding why Intel purchased Habana Labs yesterday. We are nowhere near where we need to be when it comes to the infrastructure for AI. This includes the silicon, the software, the connection between the edge device and the cloud, and the way data sets are managed and preserved so the computer can be trained.
All of this to say, we are still in the early stages of AI/ML, and we have a long road of innovation ahead from every touchpoint in the system. This is also why you should expect to see quality AI startups be snatched up quite quickly by large companies due to the competitive nature and critically importance to the future that is AI.
