Restructure Korea’s
#1 dating app home screen
Restructure Korea’s
#1 dating app home screen
OVERVIEW
OVERVIEW
Home screen redesign boosted new user activation conversion rate by 40% and paid conversion by 19%
Home screen redesign boosted new user activation conversion rate by 40% and paid conversion by 19%
The activation conversion rate for new users dropped. To fix this, the home screen was redesigned to present information clearly for quick decision-making. Key information and menus were visually prioritized, and click results were made more predictable. Two-phase A/B testing and further iterations led to a 40% increase in new user engagement rate and a 19% increase in paid conversion.
The activation conversion rate for new users dropped. To fix this, the home screen was redesigned to present information clearly for quick decision-making. Key information and menus were visually prioritized, and click results were made more predictable. Two-phase A/B testing and further iterations led to a 40% increase in new user engagement rate and a 19% increase in paid conversion.
Dec 2020 - Jun 2021
Lead Product Designer
Dec 2020 - Jun 2021
Lead Product Designer
Vision & OKR planning // User research // Hi-fidelity prototyping // Design system // Data-driven rapid iteration & A/B testing // Motion graphic design // Junior designer mentoring // Cross-functional alignment
Vision & OKR planning // User research // Hi-fidelity prototyping // Design system // Data-driven rapid iteration & A/B testing // Motion graphic design // Junior designer mentoring // Cross-functional alignment
Collaborated with
4 Engineering & 1 Data Analytics Teams
Collaborated with
4 Engineering
& 1 Data Analytics Teams
Collaborated with
4 Engineering & 1 Data Analytics Teams
Initiated project in collaboration with data analysts and project manager, designing phased a/b tests and securing stakeholder buy-in for 4 engineering teams
Initiated project in collaboration with data analysts and project manager, designing phased a/b tests and securing stakeholder buy-in for 4 engineering teams
IMPACT
IMPACT
Boosted actions of new users generated more user-to-user interactions, leading to higher paid conversion
Boosted actions of new users generated more user-to-user interactions, leading to higher paid conversion
New User Activation Conversion Rate
New User Activation
Conversion Rate
New User Activation Conversion Rate
40% ↑
40% ↑
40% ↑
Social Signaling
Social
Signaling
Social Signaling
17% ↑
17% ↑
17% ↑
Paid Conversion
19% ↑
Paid
Conversion
19% ↑
Paid Conversion
19% ↑
BACKGROUND
BACKGROUND
WIPPY is South Korea's No.1 dating app in revenue for 4 years
WIPPY is South Korea's No.1 dating app in revenue for 4 years
*Excerpt from Sensor Tower Annual Report
*Excerpt from Sensor Tower Annual Report
Launched in September 2017, WIPPY averages 200K+ monthly active users and has 10M+ downloads. Unlike Tinder, WIPPY uses a pay-per-action model with virtual currency called 'Jelly' for features like viewing profiles, sending friend requests, and starting chats.
Launched in September 2017, WIPPY averages 200K+ monthly active users and has 10M+ downloads. Unlike Tinder, WIPPY uses a pay-per-action model with virtual currency called 'Jelly' for features like viewing profiles, sending friend requests, and starting chats.
PROBLEM
PROBLEM
A 0.1% drop in daily revenue occurred due to a decrease in the new user group's activation conversion rate
A 0.1% drop in daily revenue occurred due to a decrease in the new user group's activation conversion rate
6+ month steady decline;
no apparent external factors
6+ month steady decline;
no apparent external factors
Sharp decline across metrics shows no correlation with seasonality, community response, or competitive landscape
Sharp decline across metrics shows no correlation with seasonality, community response, or competitive landscape
New user activation -20%+;
coincides with new feature rollouts
New user activation -20%+;
coincides with new feature rollouts
While existing user base remains stable, cohort analysis reveals cumulative feature additions driving decline despite individual successes
While existing user base remains stable, cohort analysis reveals cumulative feature additions driving decline despite individual successes
FUNNEL ANALYSIS
FUNNEL ANALYSIS
Only 36% new user activation: High drop-off at first screen despite smooth sign-up
Only 36% new user activation: High drop-off at first screen despite smooth sign-up
Sign-up process not the problem
Sign-up process not the problem
15+ step registration maintains 98% retention but faces significant post-sign-up drop-off.
15+ step registration maintains 98% retention but faces significant post-sign-up drop-off.
First-minute app abandonment
First-minute app abandonment
2/3 of new users quit at home screen, with 60% leaving without even a single scroll.
2/3 of new users quit at home screen, with 60% leaving without even a single scroll.
Day 1 activation crucial for retention
Day 1 activation crucial for retention
Half of non-activated Day 1 users drop off within 3 days, emphasizing first-day importance.
Half of non-activated Day 1 users drop off within 3 days, emphasizing first-day importance.
Home screen key to user activation
Home screen key to user activation
First-time user experience on the initial screen is crucial for user activation.
First-time user experience on the initial screen is crucial for user activation.
USER UNDERSTANDING
USER UNDERSTANDING
Users are hesitant to engage due to information overload, unclear navigation, and hidden core features
Users are hesitant to engage due to information overload, unclear navigation, and hidden core features
Cognitive Load
Action Uncertainty
Low Findability
Information volume exceeds single-session cognitive capacity
“I don't know where to look first... there's just too much going on”
“Love all these features but... no idea where to start”
Cognitive Load
Action Uncertainty
Low Findability
Information volume exceeds single-session cognitive capacity
“I don't know where to look first... there's just too much going on”
“Love all these features but... no idea where to start”
Cognitive
Load
Action
Uncertainty
Low
Findability
Information volume exceeds single-session cognitive capacity
“I don't know where to look first... there's just too much going on”
“Love all these features but... no idea where to start”
Cognitive
Load
Action
Uncertainty
Low
Findability
Information volume exceeds single-session cognitive capacity
“I don't know where to look first... there's just too much going on”
“Love all these features but... no idea where to start”
BEHIND THE FINDINGS
BEHIND THE FINDINGS
Structured discovery of user pain points
Structured discovery of user pain points
I led structured end-to-end research processes for in-depth interviews (Jan 2021, 12 participants) and usability tests (Feb 2021, 16 participants), from participant recruitment and screening to crafting interview protocols and usability tasks, through to insights analysis and actionable recommendations, in collaboration with PM and junior designers.
I led structured end-to-end research processes for in-depth interviews (Jan 2021, 12 participants) and usability tests (Feb 2021, 16 participants), from participant recruitment and screening to crafting interview protocols and usability tasks, through to insights analysis and actionable recommendations, in collaboration with PM and junior designers.
USER NEEDS
USER NEEDS
Users expect nearby profile matches, with profile pictures critical for checking
Users expect nearby profile matches, with profile pictures critical for checking
* Excerpt from survey (Feb 2021, 500 participants)
* Excerpt from survey (Feb 2021, 500 participants)
Nearby profile match:
WIPPY's overlooked essential
Nearby profile match:
WIPPY's overlooked essential
Despite being the most anticipated feature, over 4% of monthly VOC reports indicate users either struggle to find Nearby Profile Match or mistakenly believe it doesn't exist.
Despite being the most anticipated feature, over 4% of monthly VOC reports indicate users either struggle to find Nearby Profile Match or mistakenly believe it doesn't exist.
Vital profile photos
hidden behind UI overlays
Vital profile photos
hidden behind UI overlays
Over 1/3 of profile photos are obscured by location, age, and nickname info, hiding crucial facial features and hampering users' matching decisions.
Over 1/3 of profile photos are obscured by location, age, and nickname info, hiding crucial facial features and hampering users' matching decisions.
HOW MIGHT WE
Reduce psychological barriers while delivering clear value proposition from the first moment
A/B TEST PHASE 1
Enhance visibility by expanding the profile picture, which is user’s most crucial info
Surveys show that users prioritize profile pictures the most. However, details like username, age, and distance often cover parts of the face. B Group addressed this by enlarging the profile picture to highlight key information and ensure it isn't obscured by other elements.
DESIGN STRATEGY
Build user confidence through clear information hierarchy and predictable interactions from the first moment
A/B TEST PHASE DESIGN
Build user confidence through clear information hierarchy and predictable interactions from the first moment
With 15% of revenue generated from home screen interactions, I designed a testing approach to minimize risks while validating key behavioral hypotheses with data analysts. Phase 1 focuses on fundamental visual hierarchy improvements, while Phase 2 tests both interaction clarity and content prioritization based on user mental models identified in our ideation workshops.
IDEATION WORKSHOP
Split home screen into 'Today' and 'Play' reflecting distinct user matching preferences and engagement patterns
Ideation workshop using card sorting activities revealed two clear user mental models: passive discovery vs. active exploration. New users prefer system recommendations and nearby matches for guided experiences, while experienced users favor self-initiated matching for more control.
This behavioral pattern informed our 'Today' (system-curated) and 'Play' (user-initiated) structure, with terminology chosen to be intuitive across demographics while reducing cognitive load.
A/B TEST PHASE 2
Optimize for user expectations: Prioritize system recommendations and reduce interaction uncertainty
Building on Phase 1 insights and workshop findings, I tested both content prioritization and interaction clarity. B Group validated the 'Today' tab concept with prioritized system recommendations, while C Group added explicit interaction cues. This comprehensive approach tested both navigation structure and interaction confidence hypotheses simultaneously.
HOW MIGHT WE
Reduce psychological barriers by delivering clear navigation from the first moment
DESIGN STRATEGY
Build user confidence through clear information hierarchy and predictable interactions from the first moment
IDEATION WORKSHOP
Split home screen into 'Today' and 'Play' reflecting distinct user matching preferences and engagement patterns
Ideation workshop using card sorting activities revealed two clear user mental models: passive discovery vs. active exploration. New users prefer system recommendations and nearby matches for guided experiences, while experienced users favor self-initiated matching for more control.
This behavioral pattern informed our 'Today' (system-curated) and 'Play' (user-initiated) structure, with terminology chosen to be intuitive across demographics while reducing cognitive load.
A/B TEST PHASE DESIGN
Structure two-phase A/B test to validate core hypotheses while protecting revenue stability
With 15% of revenue generated from home screen interactions, I designed a testing approach to minimize risks while validating key behavioral hypotheses with data analysts. Phase 1 focuses on fundamental visual hierarchy improvements, while Phase 2 tests both interaction clarity and content prioritization based on user mental models identified in our ideation workshops.
A/B TEST PHASE 1
Enhance visibility by expanding the profile picture, which is user’s most crucial info
User research consistently showed profile pictures as the primary decision driver, yet UI elements obscured crucial facial features. B Group implementation prioritized visual prominence while maintaining necessary context information, testing our hypothesis that clearer first impressions would drive stronger engagement.
A/B TEST PHASE 2
Optimize for user expectations: Prioritize system recommendations and reduce interaction uncertainty
Building on Phase 1 insights and workshop findings, I tested both content prioritization and interaction clarity. B Group validated the 'Today' tab concept with prioritized system recommendations, while C Group added explicit interaction cues. This comprehensive approach tested both navigation structure and interaction confidence hypotheses simultaneously.
FURTHER ITERATION IDEAS
Learnings led to focused iterations on profile cards and Voice Talk engagement
Profile visibility enhancement
Enhancing key information visibility increased the user-to-user interaction Index by over 10%. I aimed to further improve profile cards to maximize this impact.
Voice Talk anxiety reduction
While Voice Talk clicks rose by 20%, paid conversion barely increased. Reducing uncertainty about click results eases user anxiety. Using this insight, I decided to show Voice Talk experiences visually before users try it to boost expectations.
FURTHER ITERATION 1
Expanding profile photos and adding bio previews would increase engagement by providing richer first impressions
In the end, it was unsuccessful. Fling actions increased, but social signaling decreased by 4%. User interviews showed that providing more info on profile cards satisfied curiosity too early, stopping users from taking further actions.
FURTHER ITERATION 2
Visual previews of potential conversations would reduce user anxiety
Results showed a 3.8% increase in new user trials, validating the approach. User interviews revealed that reducing anxiety and building excitement effectively drove first-time usage.
Final Output
Profile pictures were enlarged for better visibility of critical user information. The redesign prioritized system recommendations and reduced uncertainty, ensuring intuitive navigation through clear hierarchy from day one.
Visual previews of potential voice chat interactions were added to reduce user anxiety. The preview content was sped up for intuitive comprehension.
Concluding Thoughts
A key learning from this project was adapting our research methodology to overcome cultural barriers in Korea, where even people in their mid to late 20s tend to be relatively conservative about dating and stigmatize dating app users as those who should be filtered out. Through pilot interviews, I discovered how this stigma led participants to downplay their app usage and provide socially desirable responses. For the main study, I addressed this by conducting research under the guise of academic study - yielding more authentic feedback. However, the real breakthrough came through usability tests, where the 'think aloud' protocol finally enabled us to capture unfiltered user perspectives.
The academic foundation from my concurrent master's studies greatly helped establish these research protocols and build organizational buy-in for user-centered approaches, ultimately becoming the catalyst for establishing the organization's first sustainable foundation for UX research. Particularly, implementing 'VoC Read Aloud Sessions' to share raw user feedback with team members transformed initial skepticism into empathy and naturally led to problem-solving ideas.
Thanks for Watching!
Concluding Thoughts
A key learning from this project was adapting our research methodology to overcome cultural barriers in Korea, where even people in their mid to late 20s tend to be relatively conservative about dating and stigmatize dating app users as those who should be filtered out. Through pilot interviews, I discovered how this stigma led participants to downplay their app usage and provide socially desirable responses. For the main study, I addressed this by conducting research under the guise of academic study - yielding more authentic feedback. However, the real breakthrough came through usability tests, where the 'think aloud' protocol finally enabled us to capture unfiltered user perspectives.
The academic foundation from my concurrent master's studies greatly helped establish these research protocols and build organizational buy-in for user-centered approaches, ultimately becoming the catalyst for establishing the organization's first sustainable foundation for UX research. Particularly, implementing 'VoC Read Aloud Sessions' to share raw user feedback with team members transformed initial skepticism into empathy and naturally led to problem-solving ideas.
Thanks for Watching!
FURTHER ITERATION IDEAS
Learnings led to focused iterations on profile cards and Voice Talk engagement
Profile visibility enhancement
Enhancing key information visibility increased the user-to-user interaction Index by over 10%. I aimed to further improve profile cards to maximize this impact.
Voice Talk anxiety reduction
While Voice Talk clicks rose by 20%, paid conversion barely increased. Reducing uncertainty about click results eases user anxiety. Using this insight, I decided to show Voice Talk experiences visually before users try it to boost expectations.
FURTHER ITERATION 1
Expanding profile photos and adding bio previews would increase engagement by providing richer first impressions
In the end, it was unsuccessful. Fling actions increased, but social signaling decreased by 4%. User interviews showed that providing more info on profile cards satisfied curiosity too early, stopping users from taking further actions.
FURTHER ITERATION 2
Visual previews of potential conversations would reduce user anxiety
Results showed a 3.8% increase in new user trials, validating the approach. User interviews revealed that reducing anxiety and building excitement effectively drove first-time usage.
Final Output
Profile pictures were enlarged for better visibility of critical user information. The redesign prioritized system recommendations and reduced uncertainty, ensuring intuitive navigation through clear hierarchy from day one.
Visual previews of potential voice chat interactions were added to reduce user anxiety. The preview content was sped up for intuitive comprehension.
