2. Our Approach
Scan data to explore data
and find relationship
between tables
Use all 9 tables of
provided data
Use current data to show
some of the most crucial
e-commerce metrics on 4
aspects: order, product,
customer & seller
Use QuickSight to
visualize
Provide insights
Apply e-commerce
knowledge to give insights
& reasons for insights
3. Order
The market is
consistently growing
from 2017 to 2018
in both sale and order
numbers passing 20%
KPIs 4 months before
the end of year
The uptrend suddenly
dropped in September
More likely to the lack of data at this
state
4. AOV is stable
arround $125
from 2017 to 2018
This business has a set of stable high spenders in
these years
Average order value
5. Logistics
Improvement
During the first stage of this
business, it takes nearly 40 days
to get orders from sellers to
carriers, the number decreased to
a reasonable number about 3 days
in after months.
First miles delivery
In Oct 2016, the estimated delivery time was quite optimistic versus the real situation,
that is customers had to wait 20 days more than the estimated delivery time.
After that, this time has been adjusted to give more waiting space to end-buyers side.
However, it was quite pessimistic estimated time compare to delivered time, and has
no no reference value for customers
Estimated vs Real delivery time
It takes about 10 days for carriers
to deliver orders to customers
Last miles delivery
6. Cancel Orders
Correlation between canceled
orders and delivered orders over
months (applied logarithm on
canceled order)
It shows that canceled order rate
(versus successful orders) were
kept steady in 2017 and 2018
which is one of the most
successful factors in an e-
commerce business
7. Product
The health beauty category accounted for 10% of this
business's revenue
Revenue share by categories
Bed bath table and heath beauty cate took top 1 and 2
best sell categories and they continued to increase the
proportion in 2018
Order numbers share by categories
8. The product description length has been divided into
8 classes
The other side is the number of
photos of products
Product content and Sales
The fact is that most sales come
from products with only 1 to 3
photos and product descriptions
under 1000 characters
The reason for this may come from
the fact that most products do not
have many photos and long
descriptions
Or the customer's purchasing
decision does not depend too much
on the above two factors
However, it seems like using less
than 5 photos and less than 1000
characters to describe a product is
enough for buyers
9. Customer
The majority of revenue comes from
customers distributed in the east
coast region, especially Sao Paolo
and Rio de Janeiro
The fact that these are economically
developed regions in Brazil
10. Credit card payments account for the highest
proportion of nearly 80%, followed by Boleto, a typical
Brazilian payment method.
Customer Payment
11. Seller
The majority of revenue comes from
sellers distributed in the east coast
region, especially Sao Paolo and Rio
de Janeiro, just the same distribution
as customer revenue
This is a good point for logistics
because the orders will be delivered
locally faster, also it is the reason
for this seller distribution
12. The rate of positive reviews
accounted for more than 80% of the
rate per month which was a positive
sign for this business
Overall review
13. Top sellers based on review number
and their score distribution
It's a good sign to see that the top
sellers had a high proportion of
positive reviews
Top Seller Review
14. Thanks for
watching
The sudden cut of order data in the last months of 2018
makes predicting the next trend of ML model is quite
wrong. For better next time, we need to remove these
monthly data
Next improvement