- Voice app
As part of our regular hackathons at AOL, the team opted to test their skills on the new Echo Dot. With Amazon pushing their voice activated assistant into the mainstream, it seemed like the perfect time to try out creating a product in this space and be first to market.
The hackathon lasted for one week with the Engadget Briefing being the result. Operating in a loosely similar manner to a design sprint, the product went from conception to ready for launch. It became the first Alexa product within the Oath portfolio and the first tech focused content skill on the Amazon store.
With Engadget being a consumer tech focused brand with an audience of early adopters, it was the obvious candidate for launch out of our suite of brands. However, the product was developed to be scalable for use with other brands and was subsequently developed further for the Amazon Alexa Show product.
Role on project: UX lead (Oath)
We believe that by creating an Alexa skill for Engadget, we will provide a new meaningful way for consumers to receive popular content from Engadget, reinforce our brand strapline ‘for the early adopters in all of us’ and gain a valuable first to market lead. We will know this to be true when we see direct positive feedback from users, acceptance into the app store and subsequently a minimum 3.5/5 rating average.
In order to create a working product within the timeframe available, the developer in our team of three began research on the Alexa API. In tandem, design and product created the timeline, researched competitors, Amazon’s acceptance criteria, legal requirements, testing techniques for voice apps and identifying any previous voice app research by other Oath employees. Following completion, we reconvened at the end of the first day for a knowledge sharing session.
Based on research undertaken on the previous day, the team decided to create the app as a ‘briefing skill’. This would satisfy our project goal in delivering popular content to users but base its operational behaviour on user conditioning through using other functions and skills of the Alexa product. This would theoretically reduce cognitive effort and increase adoption.
Another main area of consideration was delivery of the content itself through the device. With the options of pre-recorded (initially seen as a potential USP) or text to speech available, the team opted to pursue the latter. This would remove the need to add to a stretched editorial workflow and more importantly, allow us to validate the product in the wild with as little effort as possible. Following on from the development-focused research undertaken, the team decided to use our content distribution platform to provide the most popular content which would then be fed into the Alexa JSON feed. This would also allow for international teams to easily make use of our product in the future if they wished to do so.
Prototyping and Testing
Having never worked on a voice activated product, this was a new one for me! Hearing how Amazon employees went about this in creating the original Echo, we opted to take their approach and use a small Pringles tin. Cheap, quick, low effort and you get to eat the contents after!
To set up the test, the user sat with the Pringles can in front of them whilst a voice came from behind them. In conducting tests, it was simply a case of running through the user flow designed and responding to the subject based on their voice commands and responses. This allowed us to quickly iterate the user flow and remove all dead ends.
With the skill being developed classed as a briefing skill, some of the user flow was restricted by Amazon at an OS level. In these circumstances we were looking to ensure that the right content was being served and that delivery was snappy but packing in enough content so the user felt satisfied with their experience.
The result of testing was for the briefing to be based on the five most popular articles across Engadget and for the content length to be the title and first two lines of an article.
Launch and results
The Engadget Briefing launched the following week of the hackathon and became the first tech focused content skill on the Alexa platform. Without any promotion the skill was soon picked up by Engadget fans and feedback started appearing in the app store. The majority of feedback was positive and exceeded expectations. There were however some negative comments (see below) which highlighted a fear of the team following development completion and prior to launch.
Whilst we opted for delivering content as text-to-speech early on in the hackathon, in the prototyping and testing phase we were using human voice responses. This left us unable to factor in the impact this would have until being close to launch. Whilst Amazon had done a great job in creating their text-to-speech function, to the team and those subsequently tested with, it at times had a robotic feel. Following consultation with senior stakeholders, it was decided to still launch the product and learn from the wider exposure this would bring.
At the time of writing (24/08/2017), the skill has a current rating of 3.6 out of 5, exceeding our goal set out at the start of the hackathon. With the only negative feedback being the Alexa voice, a route for the next iteration of the product is clear.
“I’ve only used this once so far, but my first impressions are good. The stories are read using the Alexa voice, it seemed quite natural…”By CJ0206 on 16 June 2017
“…glad to finally get tech news!”By Amazon Customer on 5 June 2017
Being first to market with a tech content Alexa skill, a relationship soon developed with the Amazon team. As part of the Echo Show product launch, Oath was asked to become a launch partner. Using the findings from this project and blessed with a large in-house video team at our disposal, a HuffPost video briefing was created. And of course, it features pre-recorded human speech delivery.