Shopee Email-Marketing Analytics
How to Allocate Marketing Budget Effectively?

Tools: Python, Tableau Public

This project was a final team project in RevoU. My team consisted of three people (including me). As the project leader, my responsibility was to ensure that our work met the weekly target milestones and distribute the task fairly according to members' will, capability, and availability. I contributed to the data gathering, analysis, and visualization process.

Background information

Shopee is a Singaporean multinational technology company that focuses mainly on e-commerce. It first launched in 2015 with headquarters under Sea Ltd. Shopee suffered an operational loss in 2018-2019. Although revenue continues to increase, costs & expenses have increased more significantly. One of the highest expenses is in Sales & Marketing.

​

Scope
Our analysis will be limited to the available public dataset, which is the email campaign condition in Shopee.
Dataset: Kaggle

​

Objective
Investigate whether the marketing team has sent email campaigns to the right audiences and if there was any wasted expenditure.

​

Analysis Method: Correlation-Classification Analysis

As shown in the 6th slide below, we identified some features that correlated to the user's decision to open the email campaign. One of the features showed that users who didn't open their email in the last ten days are more likely to ignore the email campaign.

We chose important features and predicted the probability (chance) of a customer opening the email campaign we sent. Then we classified them into those we will send the email campaign to (higher chance) and those we won't (lower chance).

Results

As shown in the 7th slide, Shopee can save up to 44% in expenses without losing significant audiences by targeting email delivery to a more specific audience (higher chance of opening email campaigns).

Recommendation

As shown in the 9th slide, we suggested keeping the current email marketing strategy if the value of Return On Investment (ROI) > 21. But, if the value is less than 21, we suggested revising it by targeting email delivery to a more specific audience (higher chance of opening email campaigns).

​