UXUI
UT
Agile
Camping Service Retention Improvement
/2 weeks
/2 weeks
Member
1 Developer
1 Desigenr
1 Stakeholder
1 Developer
1 Desigenr
1 Stakeholder
Tools
Figma
In depth interview, UT
Responsibility
Product Designer
Research
UIUX Design
Prototyping
User Story Planning
QA
Time frame
2 weeks
Summary
This is a project aimed at improving service activation and stickiness right after the launch of the car camping spot discovery service.
We explored ways to deliver the 'personalized recommendation' feature with limited resources.
I designed the flow and gathered user feedback through prototyping.
I designed the flow and gathered user feedback through prototyping.
As a result, the 7-day retention rate for users of the feature increased by about 20% compared to the overall average retention.
As a result, the 7-day retention rate for users of the feature increased by about 20% compared to the overall average retention.
Background & Mission
Background & Mission
Background & Mission
Camping discovery service: Improving retention through personalization
Camping discovery service: Improving retention through personalization
Camping discovery service: Improving retention through personalization
We launched GAMP, a car camping discovery service showing 20 sites by distance from user location.
We launched GAMP, a car camping discovery service showing 20 sites by distance from user location.
We launched GAMP, a car camping discovery service showing 20 sites by distance from user location.
As a new service with limited content compared to competitors, we adopted a quality-over-quantity personalization strategy—delivering customized recommendations based on user preferences.
As a new service with limited content compared to competitors, we adopted a quality-over-quantity personalization strategy—delivering customized recommendations based on user preferences.
As a new service with limited content compared to competitors, we adopted a quality-over-quantity personalization strategy—delivering customized recommendations based on user preferences.
To verify whether this strategy actually delivered value to users, we established retention metrics as our core performance indicator, and set a goal of 5% improvement.
To verify whether this strategy actually delivered value to users, we established retention metrics as our core performance indicator, and set a goal of 5% improvement.
To verify whether this strategy actually delivered value to users, we established retention metrics as our core performance indicator, and set a goal of 5% improvement.
Market Position and Service Key Flow
Market Position and Service Key Flow
Market Position and Service Key Flow
Communication
Communication
Communication
Limited Resources:
Direct User Input Over Algorithm Development
Limited Resources:
Direct User Input Over Algorithm Development
Limited Resources:
Direct User Input Over Algorithm Development
While developing a recommendation algorithm would have naturally enabled us to suggest camping spots that match user needs, implementing such logic was realistically impossible with our team of just three members. Therefore, we decided to directly ask users about their preferences and recommend camping spots based on their responses.
While developing a recommendation algorithm would have naturally enabled us to suggest camping spots that match user needs, implementing such logic was realistically impossible with our team of just three members. Therefore, we decided to directly ask users about their preferences and recommend camping spots based on their responses.
While developing a recommendation algorithm would have naturally enabled us to suggest camping spots that match user needs, implementing such logic was realistically impossible with our team of just three members. Therefore, we decided to directly ask users about their preferences and recommend camping spots based on their responses.
OPTION 01
Logic based on
collected data
+
No additional user input needed beyond exploration
-
Building the data collection logic requires significant time and resources.
OPTION 02
Collect data
by directly asking
+
Quick development with a relatively simple logic.
-
User input is needed for survey responses, but participation rates may be low.
Challenge & Research
Challenge & Research
Challenge & Research
How can we reduce drop-off rates during responses?
How can we reduce drop-off rates during responses?
How can we reduce drop-off rates during responses?
I identified concerns about user drop-off due to lengthy Q&A flows and conducted case research to reduce this friction.
I identified concerns about user drop-off due to lengthy Q&A flows and conducted case research to reduce this friction.
I identified concerns about user drop-off due to lengthy Q&A flows and conducted case research to reduce this friction.
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
MBTI test style Q&A flow of o'House
MBTI test style Q&A flow of o'House
MBTI test style Q&A flow of o'House
Approach
Approach
Approach
Fun process and anticipated results, like an MBTI test
Fun process and anticipated results, like an MBTI test
Fun process and anticipated results, like an MBTI test
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
I observed that major brands like oHouse were running MBTI-style test events that went viral. Despite having over 10 questions, users completed the entire process and shared results with each other.
Approach 01
Approach 01
Approach 01
Recommend newly updated spots campers want most
Recommend newly updated spots campers want most
According to previous interview, campers want to discover less-known camping spots. However, our existing data consisted only of popular spots since it was already exposed to users and crawled from third-party sources.
According to previous interview, campers want to discover less-known camping spots. However, our existing data consisted only of popular spots since it was already exposed to users and crawled from third-party sources.
According to previous interview, campers want to discover less-known camping spots. However, our existing data consisted only of popular spots since it was already exposed to users and crawled from third-party sources.
To address this, our team decided to personally scout and recommend newly added car camping spots. We implemented a weekly policy where users receive fresh recommendations every Friday at 6 PM as part of our new spot discovery routine to boost retention.
To address this, our team decided to personally scout and recommend newly added car camping spots. We implemented a weekly policy where users receive fresh recommendations every Friday at 6 PM as part of our new spot discovery routine to boost retention.
To address this, our team decided to personally scout and recommend newly added car camping spots. We implemented a weekly policy where users receive fresh recommendations every Friday at 6 PM as part of our new spot discovery routine to boost retention.
Approach 02
Approach 02
Approach 02
Binary choice questionnaire
Binary choice questionnaire
We designed the questionnaire in a format similar to the MBTI-style marketing events that were popular in Korea at the time, featuring situational hypothetical questions with binary choice answers. We also added a humorous nuance to make it more engaging.
We designed the questionnaire in a format similar to the MBTI-style marketing events that were popular in Korea at the time, featuring situational hypothetical questions with binary choice answers. We also added a humorous nuance to make it more engaging.
Approach 03
Approach 03
Approach 03
Turning information into rewards with a Goodie box metaphor
Turning information into rewards with a Goodie box metaphor
"To gamify the opportunity to view recommended car camping spots and make it feel more like a reward, we added a goodie box metaphor.
"To gamify the opportunity to view recommended car camping spots and make it feel more like a reward, we added a goodie box metaphor.
"To gamify the opportunity to view recommended car camping spots and make it feel more like a reward, we added a goodie box metaphor.
Design
Design
Weekly Camping Spot Goody Box
Weekly Camping Spot Goody Box
Weekly Camping Spot Goody Box
Test
Test
Test
Coupon-like Entry Point
Coupon-like Entry Point
Coupon-like Entry Point
While there were no major issues once users entered the flow, all UT participants perceived the entry button on the home screen as a coupon pack. To address this, we added a speech bubble that includes the functional definition as a camping spot recommendation box.
While there were no major issues once users entered the flow, all UT participants perceived the entry button on the home screen as a coupon pack. To address this, we added a speech bubble that includes the functional definition as a camping spot recommendation box.
While there were no major issues once users entered the flow, all UT participants perceived the entry button on the home screen as a coupon pack. To address this, we added a speech bubble that includes the functional definition as a camping spot recommendation box.
As is
To be
Result
Result
2x Retention Improvement for Recommendated Users
2x Retention Improvement for Recommendated Users
2x Retention Improvement for Recommendated Users
One month later, our 7-day retention analysis revealed that users who engaged with the feature demonstrated 30% higher retention rates than non-users.
One month later, our 7-day retention analysis revealed that users who engaged with the feature demonstrated 30% higher retention rates than non-users.
One month later, our 7-day retention analysis revealed that users who engaged with the feature demonstrated 30% higher retention rates than non-users.
Average
28%
58%
Used Cohort
Follow-up
Follow-up
Follow-up
Expanding reach to more users
Expanding reach to more users
Expanding reach to more users
We decided to launch a follow-up project to broaden our user adoption. To this end, I carried out phone interviews with the following user segments:
We decided to launch a follow-up project to broaden our user adoption. To this end, I carried out phone interviews with the following user segments:
We decided to launch a follow-up project to broaden our user adoption. To this end, I carried out phone interviews with the following user segments:
Users who used it once and then churned
Users who used it once and then churned
Users who never used the feature
Users who never used the feature
Users who continuously use the feature
Users who continuously use the feature
The results showed that churned users lost interest because the recommendations didn't match their preferences, while non-users failed to engage with the feature due to lack of awareness.
The results showed that churned users lost interest because the recommendations didn't match their preferences, while non-users failed to engage with the feature due to lack of awareness.
The results showed that churned users lost interest because the recommendations didn't match their preferences, while non-users failed to engage with the feature due to lack of awareness.
While we couldn't improve recommendation quality due to our data constraints, we changed theentry point to enhance its visibility.
While we couldn't improve recommendation quality due to our data constraints, we changed theentry point to enhance its visibility.
While we couldn't improve recommendation quality due to our data constraints, we changed theentry point to enhance its visibility.
