Thursday, 11 August 2016

Optimizing an E-Commerce site using Business Analytics and Machine Learning

In recent days many e-commerce sites are failing.The main reason has been lack of Business Intelligence and optimization on site.Many sites don't have even a simple recommendation and customer profiling.Yes, these are the part of so called business analytics and an e-commerce site can't have a long run without integrating its business model with analytics. Here are few ways of implementing simple analytics and machine learning algorithms which will definitely impact the bottom line on any organization.

1) Recommendations-

Depending on type of data we have, a recommendation engine can be built.There are multiple ways of recommending a product on e-commerce site.

A) Recommending a combo-Fitness Program+Supplement+Gear for There are certain trends and possible combos that can be put together. People are tend to buy a combo at a given discounted amount. Market Based Analysis is a famous marketing tool to find such rules.

B) Recommendation based of historic visits/purchases- There are rules regarding buying behavior of products. People are tend to buy certain products at certain intervals.Like supplements will be bought by someone over regular intervals. Data analytics can help in identifying those rules and thus increases sales.Profiling of customers helps in personalized recommendation.

C) Best Seller in multiple ways- Top seller of the day in terms of revenue, frequency of purchase, average rating, sentiment analysis of comments. People are more prone to buy already tested and recommended products by others.

D) Purchases by look-alike customers- There is a customer A with data characteristic as H1. A has not bought anything yet. There is another customer B with same data characteristics;H1, but he has bought a product P, So there are chances that if we recommend product P to A. He is likely to buy P. There are many engines are available to implement so.

E) What's trending now/What is just bought now-People are more prone to buy a product which is recently bought by other customer.

F) New Arrivals- people are interested to know- “What's new in the store?”They give more attention to these products and likely to buy more compare to rest.

2) Launching a new product/product category- 
Based on the strategy of organization, Market Gap Analysis or Conjoint Analysis can be done on product data.Based on survey data, It can be predicted which kind of product will be bought by the customers.
Many a times, CXO think that they don't have product with X,Y,Z features and launch that product. But how would they know that X,Y,Z were really needed by the customers? Market Research can help in identifying so.

3) Removing a product-  

Analytics also helps in identifying product life cycle. When should a product be taken away from catalog. Product rating and reviews are helpful in identifying a bad product. Social Media has been using sentiment analysis but e-commerce is yet to implement it.

There is an another nice article on -" How analytics can impact sales data"

No comments:

Post a Comment