Listen to TWEET
Listen to timeline, trending, and celebrities' tweets
Voice UI/UX |Work Project
Time
12.2019-05.2020
Team
1 Product Manager
2 Engineers
Responsibility
Learn user needs,
VUI design,
usability testing
Delivery
Dialogue flow
BACKGROUND
The company focuses on developing innovative voice experience, providing techniques and UX solutions of creating Voice UI for current Apps, to let users easily complete tasks like navigation, social, entertainment, food/ticket ordering by voice.
Voice Twitter
Voice Twitter is an IOS App for users getting, listening to, and composing tweets all by voice.
Listen to tweet allows users to get and listen to tweets from timeline, trending and celebrities.
Tech Process
PROJECT GOAL
Create smooth conversational experience for users
getting and listening to interested tweets.
DESIGN PROBLEM
-
What to read ?
-
How to read ?
DESIGN PROCESS
WHAT TO READ?
LEARN USER NEEDS
Technical Constraints
Twitter have tons of contents, and organizes tweets in multiple ways, like user timeline, person’s profile, news, topics, etc. The server grab information from Twitter by public Twitter API. While, It’s impossible, and no beneficial for us to implement voice flow exactly like Twitter menu, because:
-
Not all tweets group has public API to be accessed to.
-
Voice input might be obscure to target to accurate contents.
Gather User Requirements
To learn what tweets users are interested to listen to, I interviewed 20+ users by asking
“I can find your interested tweets and read for you, what tweets would you like to get? ”
I used card sorting to record answers. Queries typically fall into the following categories:
Translate User Needs Into Feasible Voice Features
By diving deeper about these categories, and collaborating with engineers learning technical feasibility, I defined the above user queries into 5 categories, which can cover almost user queries, and also be able to be implement.
HOW TO READ?
DIALOG FLOW DESIGN
DIALOG FLOW 1.0
Read 1 tweet, then stop for new queries, allow like, retweet.
With the original hypothesis that users would enjoy unique voice interaction to complete Twitter tasks, we implement the feature of like, retweet, and designed the conversation experience to encourage users' engagement.
Usability Testing
I analyzed all log conversations, and interviewed 5 users for their feedback and satisfaction.
Problems
-
Users prefer listening multiple tweets, rather than keep responding to the App.
-
Users have low will to do “like" "retweet" or "reply".
Design opportunity
-
Create experience of reading multiple tweets.
-
Lower users' efforts to command the App.
-
Focus more on "reading", rather than "interaction".
SOLUTION
We concluded that, users require the experience of the reading tweets, as continue reading multiple tweets , but keep the flexibility and convenience to stop for other queries. I ideated 2 solutions:
Ideation 1: Keep reading until user tap the Mic for a new query
Ideation 2: Read multiple tweets a time, then stop for a new query
We tested reading 3/4/5/8 tweets a time, 5 tweets a time got most user votes and satisfaction.
Decision Making
DIALOG FLOW 2.0
IMPACT
Query complete rate
92%
Customer Satisfaction Score
8.6/10
REFLECTION
Customized design: Allow users to set up reading counts
The number "5" is also a hypothesis, which need more test and research to validate.
Besides, we are thinking of provide customized setting for users to decide his own reading counts in 1 turn.