Data Analytics For An E-commerce Company

Analyzing Sales Data for Product Segments and Geographic Regions

In the present business scenario data about the customers who are purchasing products from an e-commerce site can be very useful for the business organization. Presently competition in the e-commerce business is intense and most customers have extensive range of options for a single product (Drewnowski, Michels and Leroy 2017).  Therefore, for e-commerce business organizations it is important to utilise the data analytics to improve customer experience. Organizations are nearly dissecting the purchasing way for every customer and enhancing their buying experience and making it a consistent procedure.

The following report contributes to the research methodology for a data ananlysis project, analytical findings in the project. In addition to that, recommendation based on the analytical finding and plan to implement those recommendations are also provided in the different sections of this report.

Use of the data analytics for an e-commerce data set is helpful in providing personalized experience for every customer on the website.  The organization can  enhance and personalize experience of the customers at scale. In this way they can increase conversion rates significantly. After reviewing different analytical methodologies from different datasets, the descriptive analysis methodology to find out the relation between the different elements of the dataset and predictive analysis (using the linear regression analysis) method in order to predict the monthly sales for products given in the dataset (Chambers 2017).  As the regression analysis includes handling of some independent variables in the dataset in order to find out how these variables influences a dependent variable which is in this scenario the total monthly sales of the products.  This technique helps in describing the way value of the dependent variable changes with the changing value of the independent variable.

In this e-commerce data analytics project the fundamental goals are to discover the elements in the dataset that are impacting the buying behaviour of customers and perceive the patterns and additionally designs in the selected dataset (Drewnowski, Michels and Leroy 2017). This will be helpful in predicting the buying behaviour of the customers to take care of the demand in the different regions in which the e-commerce organization is doing its business. So as to examine and analyse the sales dataset the association analysis between the different elements of the dataset is used, linear regression analysis with the specific end goal to anticipate monthly sales while considering the most influential factor.

 For the selected data set following is the first few rows of data. which shows the different kind of attributes of dataset

ProductName	ProdcutPrice	Shpping _Type	Monthly_Sales_value	Region	Customer_Number	Type_of_Customer
CheeseSauce	39	FREE	3790	QLD	27	NEW
Banana Bread	47	CustomerPaid	3092	ACT	10	NEW
Cheddar cheese	40	CustomerPaid	3333	VIC	29	EXISTING
Banana Bread	38	CustomerPaid	3310	WA	12	EXISTING
Banana Bread	31	FREE	3602	VIC	24	EXISTING

Prioritizing Products and Marketing Strategies for Different Regions

Table 1: First Five rows of the selected dataset

Region	ProdcutPrice	Monthly_Sales_value	Customer_Number
ACT	10515	1078421	6967
QLD	11152	1173713	7543
VIC	11756	1198612	8135
WA	11742	1252053	8171

Table 2: Number of customers in different region

It is evident from the above table, that the ACT region has the minimum number (6967) of customer’s respect to the total number of customers in every region.  Therefore, it can be stated that in order to improve the total revenue the organization needs to acquire new customers from the ACT region.

When the number of customers are plotted against the regions in from the dataset, the following graph is generated,

 

Figure 1: plotting of number of customers from different region

From the above table it is clearly visible that the WA regions has the maximum number of customers and ACT has the minimum number of customers.

Region	ProductName	ProdcutPrice	Monthly_Sales_value	Customer_Number
ACT	Banana Bread	2064	206844	1177
Cheddar cheese	2118	221625	1541
CheeseSauce	2009	202661	1348
Cocoa	2167	226304	1409
Flavored Noodle	2157	220987	1492
QLD	Banana Bread	2351	262903	1592
Cheddar cheese	2022	214111	1512
CheeseSauce	2742	279725	1780
Cocoa	1555	158404	1005
Flavored Noodle	2482	258570	1654
VIC	Banana Bread	2091	226953	1613
Cheddar cheese	2297	232920	1619
CheeseSauce	2264	226786	1400
Cocoa	2664	270579	1918
Flavored Noodle	2440	241374	1585
WA	Banana Bread	2370	250725	1691
Cheddar cheese	2232	237076	1517
CheeseSauce	2465	263652	1684
Cocoa	2491	276056	1738
Flavored Noodle	2184	224544	1541

Table 3: Sales of different product in different regions

For the different regions, it is important to prioritize the products according to their sales. As one-template for all is not helpful in this scenario. Therefore, from the above table its can be stated that, For ACT region the least sold product is, CheeseSauce for which the total sales are given by the figure, $202661. For the Cocoa is the least sold product in the QLD region which generated the total monthly sales for $158404.  for VIC region, the least sold product is again CheeseSauce with the total monthly sales $226786. At the end for the WA region it is observed that, Flavored Noodle is the least sold product with the total monthly sales $224544. Therefore, these products should be marketed properly and prioritized in order to increase revenue from the business as well as business performance in different regions.

In addition to that, the sale of the products in different regions are plotted which is given below,

 

Figure 2: graph of sale of different products

From the above chart it is clear that, the highest banana bread is sold in the WA region, for Cheddar cheese the region is VIC. In addition to that, most number of orders for CheeseSauce, Cocoa, Flavored Noodle are from the regions QLD, VIC and QLD regions respectively.

When the number of the customer paid and the free shipped products in the selected dataset is investigated it is found that the, in every region the number of customers as well as total monthly sales for freely shipped products are higher than the customer paid shipping orders.  Therefore, it is suggested to provide free shipping for the products that will easily attract and influence the customers from the different regions where they e-commerce operates.

Shipping _Type	Region	Monthly_Sales_value	Customer_Number
Customer Paid	ACT	533972	3473
QLD	538437	3580
VIC	567153	3854
WA	633082	3981
FREE	ACT	544449	3494
QLD	635276	3963
VIC	631459	4281
WA	618971	4190

Predicting Monthly Sales using Linear Regression Analysis

In order to predict the monthly sales from the given data set with the change in the number of customers for order provided in the dataset.

Here it can be stated that, when the number of customers for an order gets increased than 25, the monthly sales of the product can reach up to $6000.   The result of  the regression analysis is provided in the following plot,

 

Figure 3: Linear regression prediction model for monthly sales prediction

Providing free shipping for all the products: As the number of freely shipped products larger than customer paid shipping products therefore it is suggested to provide free shipping for all the products.

Customized product recommendations for the customers: There are various products that are purchased together by the customers. Along these lines utilization of this items to influence different other customers with the goal that they can purchase the set of the same products prompting the enhanced business and also revenue from the areas like ACT and QLD region.

As the preferences of the customers are distinct therefore, it is suggested to customize the home pages of the E-commerce site according to the preferences of the users after they login to their respective accounts.

The analysed data about the customers can be used as a driver in order to customize the product pages in order to attract the target group of customers on the website.  On the contrary of the traditional way the customers have to go through multiple pages on the website to find out the required product as the home pages are designed with the one-for-all template for the home pages to win customers.

Therefore, it is suggested to take extra step in order to attract the customer by showcasing very customized products on the home page itself and then encourage the customers to indulge them into immediate buying by promoting the previously purchased products.

As the number of freely shipped products are more than the customer paid orders therefore it is suggested to provide free shipping for all its products.   In order to mitigate the issue of decreasing profit due to the free shipping products it is suggested to offer ‘free shipping’ for the products with minimum item numbers or minimum order amounts.

This would be helpful for the organization in order to drive usual order value as well as more profits against which the invested shipping cost can regulated.

Moreover, ‘free shipping’ campaigns for the products on the ecommerce website can improve the conversion rates (from visitor to loyal customer) to a considerable range for the organization compared to the other organizations that does not provide free shipping for the products.

Conclusion 

Using liner regression analysis technique to design the prediction model can be valuable for the e-commerce organization as it will assist the organization with exploring patterns among the distinctive components of dataset identify with the customers purchasing behaviour of an region. With the analysis of the dataset the it is evident that the organization should try to acquire new customers and changes in prioritizing the products in different regions to improve business and revenue from it.

References  

Chambers, J.M., 2017. Graphical Methods for Data Analysis: 0. Chapman and Hall/CRC.

Drewnowski, A., Michels, S. and Leroy, D., 2017. The impact of Crunchy Wednesdays on Happy Meal fruit orders: analysis of sales data in France, 2009–2013. Journal of nutrition education and behavior, 49(3), pp.236-240.

Rink, J., SAP SE, 2015. System and Method for Apparel Size Suggestion Based on Sales Transaction Data Analysis. U.S. Patent Application 14/231,587.

Van Donselaar, K.H., Peters, J., De Jong, A. and Broekmeulen, R.A.C.M., 2016. Analysis and forecasting of demand during promotions for perishable items. International Journal of Production Economics, 172, pp.65-75.