Let's imagine you turn off your kitchen lights everyday... But one day you forget about it before going to bed
Alexa will do it for you, and you can sleep in peace!
‍
That’s a proactive action (aka Hunches); Actions that are triggered without the user’s input

In order to enable Alexa Hunches to learn and be more efficient, a feedback system is needed.
‍
In a course called Studio Practice, through industry mentorship, our team has redesigned the current feedback system for proactive actions to achieve transparent communication, to reduce cognitive load caused by overly complicated system, while raising awareness about the available features.

Picture that you sleep at 10 pm every night... But your project deadline got you up till 12 am... Alexa turned off the light at 10:15PM. Based on your routine, Alexa assumed you went to bed and forgot to turn off the light. Therefore, she intervened by turning off the light, which was an incorrect action.

Providing feedback to Alexa leads to fewer incorrect actions, which Improve user’s trust.
While there is a current feedback system for Hunches, we had identified the following problems.

Problem 1 (P1): Lack of user awareness about Hunches and its settings.
Hunches notifications are turned on/off for all devices. They are off by default, and by most users.
Problem 2 (P2): The app does not give sufficient context regarding the action taken
Problem 3 (P3): No measurement for the impact of feedback given

Our solution is catered towards those, who use the Amazon Alexa ecosystem to control at least one smart home device (eg. lights, smart locks, thermostats)

Research Insight: Users prefer to be notified in a semi-proactive manner.
‍
By leveraging the Echo Show's always-on display, we can grab the user's attention to notify them of proactive actions triggered by Alexa without being intrusive. From there, they can reach the activity page on demand, by a click.

By enhancing Alexa's widget, we integrated hunch notifications to deliver updates on the user's home screen without intrusion

‍Research Insight: Intrusive notifications are preferred during high-stakes situation
‍
While enabling devices for Alexa's hunches, users can choose to set alert types for each selected device without any additional steps

In the current version, Activities are hidden under settings, and it requires the user to go down 4 levels in the hierarchy. This makes proactive actions hard to find.
‍
In our version, a new "Hunches" section appears on the home screen to alert users of actions initiated by Alexa. Additionally, a new tag is introduced in the device tile will inform users of proactive actions performed on it.

We challenged ourself. Should feedback be limited to yes or no? How can we gather more contextual information from user feedback?

While context dependent events are predicted by pre-defined parameters (eg. wifi, location, calendar), they cannot reduce the fault-rate of Hunches over time. Preference dependent event are predicted by usage patterns, and they can reduce errors over time.
‍
Disclaimer: This was based on our logical assumption that more detailed feedback would help with reducing errors

‍Research Insight: Users were more likely to give feedback when the impact of feedback was transparent
Demonstrating Alexa's hunch model enhancements via iterative user feedback to highlight accuracy improvements

Demonstrating Alexa's hunch model enhancements via iterative user feedback to highlight accuracy improvements

Being a graduate project, our solutions have not gone into production, so assessing the real impact is not possible, though we expected the followings:

Improved accuracy of Hunches, by providing contextual information through feedback

Heightened customer awareness about proactive actions

Lessened cognitive load on setting up Hunches preferences (eg. place in app, notifications, priority)

Promoting trust in Alexa, by showing the impact of user feedback, which may lead to customer retention