Team
TikTok Design, Global Community Product
Role
Product design intern
Collaborators
1 Product Manager
1 Content Designer
1 Front-End Engineer
1 Data Analyst
Timeline
1 week
My Contribution
As for my first project during the internship, I quickly ramp up on the context of Lemon9 and SEO as well as design and collaboration workflow. Within a tight timeline, I delivered high standard design solutions in one week.
In the process, I contributed data-driven insights to the product team, and proposed experiment group set up for the A/B testing. Collaborating closely with PM and DA, I analyzed testing results in depth, and provided roadmap for future optimization.
1 week
agile iteration
+2.3%
DNU conversion
end-to-end
design ownership
ABOUT SEO POSTS
Driving organic user growth at a low cost
SEO post websites provide users with gated experience and guide them to download the APP for more. Users discover Lemon8 contents on search engines, and continue to view more through downloading Lemon8 APP. As Lemon8 is at the critical stage of rapid user growth, SEO can drive sustainable, organic DNU growth at a low cost while also enhancing brand visibility and long-term advertising value.
CHALLENGE
Desipte traffic growth, video posts' new user conversion tends to decline
After Google's algorithm adjustment and Lemons' traffic optimization projects in early July, mobile traffic for video posts continued to rise.
However, Daily New User (DNU) conversion rate for video posts was lower than article pages and continued to decline, requiring urgent optimization.
Compared to photo posts, Videos have lower first-screen efficiency. The video frame takes a larger space, thus it requires active scrolling to learn about captions.
This led to lower exposure of core acquisition modules, which could then led to lower conversion.
Photo Posts
Video Posts
LEARN FROM PAST EXPERIMENTS
TikTok boosted DNU acquisition by downplaying the comments section
TikTok Landing Page
Key Action: Compress video frame height, remove related search results and comment section.
Results: Removing the comments module increased overall DNU conversion by ▇▇▇▇.
➡️
TikTok SEO Page
Key Action: Collapse the comments module to the interaction button; clicking it displays a floating layer.
Results: Global DC DNU + ▇▇▇, DAU revenue + ▇▇▇.
before
after
DATA ANALYSIS
Unmatched visual emphasis and module importance
I offered to analyze conversion rates of each consumption block module, grounding design with data, and found modules with much less contribution (e.g. Logo, author, and engage bar) occupy the top areas and have higher exposure. This dilutes exposure to more important acquisition positions like "related posts".
Meanwhile, comments section in photo posts contribute higher compared to video posts, while media area in video posts contribute a lot more than that in photo posts.
Ranking for DNU contribution:
Photo: related posts > read more >> photo area >> comment >> engage > profile > logo
Video: related posts > video area ≈ read more >> comment > engage > profile > logo
🤔
How to re-allocate visual emphasis to increase exposure of modules with higher DNU contribution, thus bringing DNU growth?
SUCCESS METRICS
Core Metrics
Video posts' increase in DNU acquisition by Daily Count
Photo posts' increase in DNU acquisition by Daily Count
Overall DNU acquisition efficiency
Monitoring Metrics
Consumption metrics for the acquired new users
User retention by 7 / 30 days
Duration users stay on SEO posts
VIDEO POSTS DESIGN SOLUTION
For video posts: Compress engage & author, comments, and captions
As video posts require a more urgent optimization, I focused on the design opportunities for video posts first. Based on the researches, I proposed the following directions to the team:
Although in the initial discussions, PM inclined toward exploring A & B only, I proposed C, which could display "related posts" section without requiring any scrolling of the user and thus increase its exposure most aggressively. Due to limited traffic of SEO, we could not include too many experiment groups, and decided to explore D in future iterations.
I proposed the A/B testing set-up as follows:
… and delivered clear and detailed UI adaption for engineers:
Comments panel for V2
Captions for V3
PHOTO POSTS DESIGN SOLUTION
Compress comments to increase exposure for "related posts"
As the comments section contributed more for article posts compared to video posts, we took more cautious steps in designing the experiment groups:
A/B Testing Insights
In the A/B test, we observed the best results in v2 of photo posts and v3 of video posts, and proceeded with these design options.
We took a deeper look at the performance of V3 for video posts. The greater loss in "read more" module likely attributed to the blank space in video frame. Due to limitations in experiment group number, we kept the video adaption rule unchanged, which compromised the caption reading experience in V3.
The increased user acquisition in media area, which we had not anticipated, suggested that aligning the video area layout more closely with the standard immersive video feed contributed to the observed improvements. Given this, improving the video viewing experience could be a promising direction for further iterations.
FUTURE ITERATION
Based on analysis on the A/B testing results, I proposed future iteration directions for video posts.
TAKEAWAY
Product designer as a strategic partner
In this project, our PM reached out to me with a detailed PRD including rather specific design solutions. Instead of just following the given path and doing the visual work, I complemented the research insights with a data-driven approach from a designer's perspective, which in the end allowed us to arrive at a better plan. This reinforced my belief that product designers' irreplaceable value lies in actively shaping strategy rather than passively executing it.
