Business Problem:
During my time at Mercari, I noticed there were a bunch of high-grossing categories that were collectibles. If this made up a high percentage of GMV then it would be business savvy to have a better understanding of the buyers and even sellers in these categories. Over at eBay, there are managers for each item category that know specific release dates for collectibles or when they would be reduced in price. This ties closely to Customer Relationship Management which helps the business set metrics on how well they are taking care of their customers/user base. Given that not every user has the same relationship to the business, best thing to do is to determine which metrics bring the most success to the business. And for users that have highest score for each metric, push the most engaged sellers for example, to the forefront of the market (give their listed items top rankings in search impressions), match them with buyers who are frequent purchasers and influence the users who are not as engaged.
So let us go ahead and determine who is thriving in our marketplace (and give them incentives) and whether or not the community itself is thriving.
Metric to Lift:
Engagement metrics (specific to area in a funnel as close to bottom of funnel/conversion as much as possible)
Hypothesis:
Those who are highly engaged are evangelists whereas those that are more casual will less likely help spread brand awareness and bring others on to platform.
Analysis Approach:
a. Examine products and see which are highest grossing b. Set metrics for customers - Recency, Frequency and Monetary c. Split customers into different groups based on metrics - RFM Segmentation, Cohort Analysis d. Combine metrics to determine health of community - Customer Lifetime Value
Code
Customer Relationship Management Python Code
Recommendation to Business Stakeholders:
Top 3 RFM Segments are: Hibernating, Loyal Customers and Champions. These are based on Recency and Frequency Metrics that measure how recent their purchase were and frequency are their purchases. Given that the top group is hibernating (25%) this means there needs to be incentives to help awaken those that could churn. Though what is comforting is that Loyal Customers make up 18% of the user base and Champions 16%.
Also, when looking at cohorts’ retention over time based on purchase time, there can be less of a drop off at each period, if buyers felt they were getting a bargain. Hence, subsidizing payments would help until there is more stickiness.
Lastly, when looking at Custmer Lifetime Value, we can see which customers we need to invest in more or encourage to bring others on the platform based on their predicted number of purchases over a given week or month. CustomerID’s: 13408 & 18102 need more love and care or convincing whereas 12748 & 14911 would be the evangelists.
Conclusion:
CRM helps us see how well business is doing in people’s everyday lives. If there are customers that are durable ones across up’s and down’s of economy, these are the ones that the business wants to invest in and create more like them.
Reference:
Harvard Business Review (HBR)