Twitch Impressions Extension
Who:
Myself (John Scott Siri)
Peter Eskander
Peter Jough
What: Based on a prior study done, we designed and tested a Twitch extension for getting higher quality feedback from users.
When: Fall 2019
Where: Newark, NJ
Who:
Myself (John Scott Siri)
Peter Eskander
Peter Jough
What: Based on a prior study done, we designed and tested a Twitch extension for getting higher quality feedback from users.
When: Fall 2019
Where: Newark, NJ
Twitch is a livestreaming service where users with other users streamers while watching streamers do different activities, often with their audience. Inspired by a previous study, our team designed a tool to help bridge the communication gap between streamers and their viewers through user-triggered surveys that could be accessed and received by streamers.
Over the course of 3 months, we used user-centric methods combining user research, design practices & wireframing, and user feedback to iterate and improve the design.
Our early drafts were based around the idea that an inclusive tool to help convey depth and simplicity user feedback. We started designing on mobile first.
While initially concerned primarily with quantitative data, we quickly incorporated the qualitative layer because we believed it added value to both the user and the streamer.
We did not want the product to be dependent on payments, but also did not want to facilitate spam. This led to us trying to create scarcity with the impressions system by restricting its usage to one per hour.
Prior to the interviews, we performed guerilla tests from more than 10 co-workers in our lab, all Twitch users. We incorporated their bits of feedback into the design before testing.
We conducted 5 speak-aloud tests from Twitch users, followed by interviews. We gathered a few insights:
1. The wireframe took 6 clicks to navigate.
This was fine for a starting point, but we were aiming for simple and this indicated a need to keep it straightforward. Following user feedback, we managed to reduce it to 3 clicks.
2. 3/5 participants said that the UI didn't indicate its functions well.
In particular, they did not feel like the icons (including the impressions icon) matched and felt like the placement of the elements did not feel satisfying.
3. The general impression was positive.
"I think fundamentally this is good because I can
share more explicitly how I feel. This just needs
to simplified more." (M, 23)
"It was good, you could provide feedback without
closing/blocking the stream." (M, 27)
We incorporated our feedback into the design and came up with an improved version of the mobile application. However, participants mentioned that the Twitch app did not have the full span of functionality that the desktop version had. Following this, we turned our attention towards desktop because it is a richer user experience with the full set of Twitch's functionalities.This led to a general reformatting of the application features.
We also wanted to identify if the participants thought that it brought value to the streamers. In order to test this, we also designed an analytics portion.
After doing this next iteration of wireframes, we opted to do another round of user tests.
We conducted 5 more speak-aloud tests from Twitch users, followed by interviews. This time we had a few more insights regarding improvements:
1. The font was too difficult to read on large screens.
A couple participants mentioned that they needed to make the prompt box bigger and had trouble deciphering tracking words on a glance.
2. Participants struggled to find the Impressions button.
The participants did not like the original location of the Impressions button, which was on the video player. This was not intuitive given that the chat box was used for most forms of stream engagement.
3. The analytics section needed to explain itself better.
A few parts of the analytics section needed to label themselves better. The need for data clarity was not readily apparent without these labels, so users were confused.
"It's hard to read the font, it's too small, make the prompt box bigger." (M, 24)
"The buttons on the video player screen are meant for settings, surveys don't go there." (M, 22)
"Also, the Y axis has no number. There's a lack if clarity for what these words mean." (M, 21)
While design is never truly finalized, this is the final iteration of the project given the feedback we received within the project timeline. The Impressions tool has demonstrated its fair share of potential within a nuanced system and was generally agreed by participants that it would likely be beneficial in practice.
Given that viewers believe that their emotional support and feedback makes a difference, and the Impressions tool makes it clearer. This is beneficial to viewers wanting to express themselves and to streamers looking to keep their audience satisfied.
