The LINE sticker official account, which has many friends (more than 73 million) even amongst LINE official accounts, sends numerous messages to countless people every day. To increase usability while also adhering to the restrictions placed on the number of messages that can be sent following the recent platform policy changes, a more efficient way of sending messages was needed.
As a result, we improved the message sending by creating scores that predict the probability of checking and clicking messages using properties from LINE official accounts, friends, and relationships. This has doubled the number of user clicks compared to sending traditional messages, tripled the sales of stickers via messages from LINE official accounts, and improved the sending efficiency.
This session introduces an example of improving the efficiency of sending LINE official account messages from LINE stickers using machine learning and data science.
