


The selfie has been integrated into modern life as a significant part of social media. BeautyPlus is committed to enable an enjoyable selfie experience, including one-button selfie-taken, one-button beautification. As a large part of the market taker, BeautyPlus has to explore and apply AR technology to enhance user experience.
Early insight from user research

At the outset of the project, we didn't have a clear clue of how much users will accept AR selfies. To better understand the localized perception and preference of AR features in different countries, I partnered with the marketing team to explore the usage of AR in the countries where we operate.
We used an in-app survey, social channels, and field research to collect 19762 samples of individuals, most of which are from Japan, India, and Indonesia.
Discover
Reason why users don't like to use AR filter
We found almost 30% of users don't want to use AR filters, and there are two main reasons: 14% of users prefer to use traditional selfies rather than new AR filter. 12% of people don't know what AR selfie is.

Problems about using AR feature
We collected data from the other 69% of people who usually used AR function and tried to find pain points when taking an AR selfie. 35.8% of users have trouble with finding AR filters they want.

Deep insights
To the deep understanding of the challenge users faced to use AR selfie, I tested the existing Beautyplus app with 5 participants and monitored the whole process. I was so surprised that three people spent over 5 min to find AR function. 2 people said it was still hard to see this feature.
Our data also shows that only 34.6% of users enter the AR selfie page. That means we need to optimize the current AR function icon to give a more clear guide.

AR filters in different countries
Cooperated with the data team, I created this chart to represent the data performance about the single AR filter in the different countries. Here are three simples.
It is evident that "White bunny" and "Pink Cat" are very popular in Korea and Japan, but the data in the US, Brazil, and India, it's not good as we expect. On the opposite, "3D brown bunny" has a high performance in the western countries, but not very popular in Asia.

AR filters ranking
I listed TOP 10 AR filters in the US and Korea to know the preference of AR-style.
US
Korea






Refraining the problem
1. The AR icon on the selfie page is not attractive enough.
2. "Big different preferences in different countries."One AR can't be accepted in all countries. Realistic style and 3D modeling are popular in the US and other western countries. On the other hand, Eastern counties like Japan and Korea prefer to use the 2D cute drawing style.
3. Need more explicit guidance to tell users what the AR selfie is and how to use it.
4. Reorganized all AR filters that easy for users to find.
Solution
1. AR icon optimization


2. Promote different AR filters in different countries
Optimize the "Hot category" is an excellent way to promote AR filters in different countries. Based on culture preference, we can put a localized AR filter in the hot category.
US
Korea


3.Automatically open AR feature
Achievement: The CR increase 5.2% on IOS and 8.9% on Andriod
"Don't assure that everyone plays AR before." Show what AR is in the first place. Over 65% of users get into the Selfie feature, but only less than 30% of users would like to download AR features. My solution is automatic open our most popular AR sticker when the user clicks the AR button.


4. Set up “Categories”
Achievement: The number of "AR photo taken per user" increase from less 5.0 to over 8.0
Based on the research, "Hard to find certain AR filters" is a big problem. My solution is creating categories based on the AR element. I created groups like HOT, Glasses, flower crown, and animals. It's easy for users to find what kind of styles they want to use.


5. Design "characters"for AR thumbnails
Some of AR filters have an excellent CR(Conversion Rate)over 50%. It shows that when users open and use these AR, they love it! But the total number of downloads is deficient. This situation represents there may some issue "stop" users before opening the AR filter. It's a perfect way to increase the download of each filter. For example:

