Amazon Echo Dot

Anticipatory Design proposal to Conversational Interface

Ux Research
Usability Test
A/B Test
After observe the importance of voice user interfaces and personal assistants like Siri, Google Assistant and Alexa. I decided make a academic research about this for my undergraduate thesis. Between a lot of assistants, I choosed Alexa - specifically the Amazon Echo Dot device. Because I saw the opportunity to propose something new and why not, increase your competitive at industry.

Before start my research and understand how the assistant works, I bought an Amazon Echo Dot and began using it everyday with the main objective undestand if personal assistant really improve the productivity or if is just a new gadget. During the experiment period was evident that virtual assistant always needed a command to provide some information to me. After that I thought to do something easier for the users.  

During my research, I found some academic papers about Anticipatory Design. Basically, it works to understanding user behaviours, collect data about yours routine and anticipate decisions.

So, after understand how Anticipatory Design works, the next step was create tasks for usability tests. To keep the consistency between the tests, I made an A/B test for the users choose whats the best version between the current version (Proposal A) or my proposal following Anticipatory Design principles (Proposal B).

Five users were recruited for the usability tests. Afte each section, I applyed the System Usability Scale (SUS) method to collect metrics, and validate which version wad better acceptance according the user preferences.
Fiver Users
Recruited
Proposal B
Most accepted with
68 points
User feedbacks
Positive interest for Proposal B because the humanized interaction.
Anticipatory Design can be the differential for the future of conversational assistants.
If you have interest know more details about the research. Check the academic paper avaliable at Interaction Latin America 2018
Design Workshop
Turbogram Redesign