Find Segments (Post Feedback)
Segment merchants based on their two year transaction history
Created on October 6|Last edited on October 6
Comment
The dataset given has 3 separate fields:
- Merchant ID as 'merchant'
- Time of transaction as 'time'
- Transaction amount in cents 'amount_usd_in_cents'
Segments
Used K-Means clustering on a lot of normalized data rolled up at the merchant level.
Key features ended being used for the segmentation:
- Lifespan in days (log transformed)
- Number of payments over lifespan (log transformed)
- Average order value (log transformed)
We found 3 well separated clusters

Projected 2D views



What do these Segments mean?

How are these segments different from each other?
Purple segment is slightly smaller than the rest, but otherwise they're all pretty evenly distributed segments.
Long Time Loyalists (Purple)
- These merchants have spent an order of magnitude higher with us than the rest with an average LTV of 54.2K Dollars
- Their average lifespan is 427 days
- Their orders though are similar in size to the rest at around 14 dollars per order
High Frequency Short Termers (Green)
- These customers spend 2.5 times per day
- Their order sizes are comparable to the long time loyalists
- But inspite of their high frequency and good order sizes, they churn quickly, with an average lifespan of only 17 days
Slow Movers (Yellow)
- These merchants purchase only once in ten days, an order of magnitude lower than the other group groups
- But their Average order values are 50%-70% higher than the other two groups
What does it mean for the business?
Long Time Loyalists (Purple):
- These merchants keep the lights on.
- Each merchant in this segment is almost 9 times as valuable as an average merchant from the other two segments.
- They are loyal and make recurring purchases at frequent intervals.
- They also spend considerably higher time with us than the rest
- Identify the reasons behind their high affinity towards our product.
- The reasons could be service, products, specific features etc.
- It's possible that these merchants would enjoy a more regular payment strategy - i.e. a subscription service to avail of our services as evidenced by their loyalty and affinity
High Frequency Short Termers (Green):
- These customers although make 2.5 times more purchases per day than the long term loyalists, they spend only 1/30th as much time as the loyalists on average
- We should understand the reasons behind their high churn
- Since it is unlikely that they would be churning after good comparable sized orders and rapid purchases, it's likely that their needs are being met rather quickly and they don't need our services for longer
- Identify what can be done for them to use our services for a longer interval, or if we could create a product that retains most of their highly used functions, but add additional features that they find valuable
High Potential (Yellow):
- Since their AOVs are similar to our loyalists and lifespan is also a good 200+ days, it is likely that they value our product
- But they value our product only a tenth of what the loyalists do, and hence the much lower frequency
- It is possible that the service we provide them is only a part of all the services they use, and that identifying what services we can offer to keep these high value customers coming back and using our services more frequently, would likely significantly add to the overall revenue, given the harder part - new customer acquisition - is already complete.
- Increasing their frequency to even half as the loyalists to once every five days on average would raise the per merchant LTV by 400 times in that group.
Add a comment