Enhance experience and boost data performance by refining information architecture
Product designer / 2018-2019
Weibo, the leading Chinese social media platform, implemented a revision in 2018 after a year of A/B testing experiments on its app.
Our home feed needed a boost in refresh and exposure, to offset the drop of data caused by adding the Videos tab to the bottom navigation.
I led the IxD for Weibo App new revision, it was another story.
To improve long-term data performance, as a staff designer in the Feed Team, I decided to restructure the Weibo post architecture to enhance the overall experience of browsing, aiming for deeper user engagement.
Weibo Posts are the primary way users share updates, images, videos, and links. They are central to communication, user engagement, information dissemination, real-time updates, and social interaction.
Posts serve different purposes in various scenarios
These key scenarios make Weibo a versatile platform, catering to a wide range of user needs:
Pain points
1. The number of hierarchical layers
in our posts is overwhelming, creating a reading burden;
2. Too many redundant content attached by us (Weibo as first party), while it lacks reasonable guidelines;
3. More than half of the posts are
longer than one screen on a phone, making reading inefficient.
Break down post structure
Display Window&Tag: direct to content related products; Info: display data and status.
Key findings: None of the competing products automatically match first-party traffic-directed cards or hashtags to user-generated content as Weibo does (except ads). Additionally, the media display rules of best practice products are more reasonable than Weibo.
Video View data:
• Fb and Instagram’s VV data are showed below the video.
• Twitter and Qzone’s VV data are showed on the video.
• Fb shows the VV data only when met a certain threshold.
Exposure/reading data:
• Q-zone showed it under the content, while all other competing products do not.
Username + Location / Group / Activity / Feeling
– Content
1. Number of likes / scrolling friends + number of comments, shares and video views
2. Like, comment, share (public content) buttons
3. popular comments
4. comment box (after medium been viewed)
Name @Username / Ad label
– Content (w/video length, ads can come with title + button)
1. Comment, retweet, like, view data, save, share buttons (rolling data updates)
Username + Location
– Content
1. Like, comment, share, save buttons
2. View data, who liked – Caption
3. Number of comments + popular comments (1 displayed)
4.Comment box (after stays for 3s)
5. Post time, translate button
Q-zone
Username + Post time
– Content (w/ view data, length on video)
1. Album
2. Upload device type
3. Location, date
4. Number of views; Likes, comments, repost buttons
5. Number of of likes
6. Number of repost (username)
7. All comments
8. Comment box
WeChat Moments
Username
– Content
1. Location
2. Users mentioned
3. Post time, action button
4. Mutual friends’ comments
After organizing and clarifying the information hierarchy of our posts, I developed multiple strategies to shorten their length, aiming to boost exposure, increase refresh rates, and enhance post consumption.
Strategy details
• Clarify: Display similar information together; Separate clickable areas and non-clickable data in different rows.
Tags and reactions are clickable, while video views is not.
• Simplify: prioritize user-generated content and tags. Downgrade and weaken Weibo first-party matched tags or cards when there are too many information hierarchies.
Prioritizing user-initiated elements; If exist, downgrade auto-matched window to a tag.
• Develop rules:
- Display Window/Tag: User can post with windows and location tag, additionally, Weibo can add one auto-matched window or tag; Prioritizing user-initiated ones over auto-matched.
1. User-posted window or location tag: auto-matches a tag or window respectively.
2. User posted both window and location tag: no auto-matching.
3. Remove attached-cards (except ads), use window, and it must be displayed separately from media.
- Popular Reposts and Likes: Only display in specific scenarios, without showing with auto-matching windows and tags.
Further iteration
Prioritize high-value info: In the 'For You' feed, where user interests outweigh timeliness, posting source such as time and device type will be replaced with author identity, recommend reasons, or super groups.
Remove 'hats': Some posts have 'hat' with sharing range and attributes, like recently followed or close friends. To simplify the structure, I chose to move these titles to the source area, beneath the username.
Details & comparisons
While developing new norms to optimize the information structure of posts, which serve as the primary content framework for Weibo, I communicated, presented, and coordinated with various stakeholders.
My design strategies yielded positive data results in the gray scale testing phase, and upon full release, achieved the initial goal:
Team:
Me, 1 designer, 2 product manager, 4 developers/engineers, 1 QA
— Weibo Post Structure Upgrade —
That’s it, thank you!
by Akah (Yaxiong Fu)