=--

A Project Report

On

“A study on Recommender Systems employed by Indian e-Commerce companies”

By

Updesh Singh Chauhan

PGDM No. 19057

Submitted in partial fulfillment of the requirement of

Summer Internship Programme

Under the Guidance of

Mohamed Minhaj.

Associate Professor – Systems,

Chairman - Campus Systems

SDMIMD, Mysore
SDM Institute for Management Development

Mysore, Karnataka, India

June 2020
 CERTIFICATE BY FACULTY GUIDE

SDM INSTITUTE FOR MANAGEMENT DEVELOPMENT, MYSORE

This is to certify that Roll No 19057, Updesh Singh Chauhan of PGDM batch
2019-21 has satisfactorily completed the Summer Internship Project titled “A
study on Recommender Systems employed by Indian e-Commerce
companies” from 8th September 2020 to 8th November 2020 to partially
fulfill the requirements of the PGDM program under my guidance.

Signature: Date:

Name:

Designation:

SDM Institute for Management Development, Mysore

Institutional Stamp
 ACKNOWLEDGEMENT

I sincerely thank Shri DharmasThala Manjunatheshwara Institute of Management

Development, Mysuru for providing me the opportunity to undertake this enriching project in
the field of Marketing Analytics for two months as my Summer Internship Program. It was a
great learning experience, and the project has given me a background of strong learning
which would be of great help to me in the future.

I would like to take this opportunity to express my sincere thanks and warm greetings to my
project guide, Professor Mohamed Minhaj, for his constant guidance and assistance.

This project would not have materialized without his constant mentorship and
encouragement. His guidance and valuable inputs at every step during this project have
helped me complete the project successfully. I am profoundly grateful to him for sharing his
truthful and enlightening views on several issues concerning the project.
 EXECUTIVE SUMMARY

The project deals with the study of the recommendation system employed by Indian
originated E-Commerce companies- Flipkart, Myntra,Snapdeal, and Big Basket. The Project
begins with exploratory study of papers and articles over the past three years. The project
makes use of filtering techniques at different situations perceived by customers, The
Research presents an overview of recommender systems that covers the framework and
approaches of the recommendation models, recommender system through up-sell and cross-
sell, recommendation engine, recommendation algorithm, methods for recommendation
system and various qualitative aspects of the selected companies. Initially, The project begins
with an introduction to the Indian e-commerce industry and company profile of the selected
companies, which are among the leading but as per current scenario only Myntra has been
taken over by flipkart, and Snapdeal having lesser market share are no longer to the
competition. So, The recommendation system study continued with selected two indian
companies i.e Flipkart and Big Basket

The analysis begins with studying the basic introduction of the companies, and further
explaining “why and what” of the selected companies. The study further continued with the
role of data analytics and recommendation algorithm techniques used by the selected
companies.Next, Elaborating a brief about why the particular technique used with several
examples.Finally a comparative study has been done of the selected companies and Next the
competitive factors and external and internal analysis using Porter’s Five forces and SWOT
analysis of the selected companies.

From the analysis, I could understand that Flipkart has been more effective and leading in
using recommendation systems compared to other Indian E-commerce companies. Flipkart
has also outperformed Snapdeal and other indian e-commerce companies in terms of market
share, diversification, market size,Profits and revenues. The Two selected companies employ
Content based recommendations systems and Hybrid of Content and collaborative
recommendation techniques.Since these companies have different product segments and also
different purchase behaviour history which lead to have different algorithms on the
recommendation engine. In addition to that, A new technology to be developed at their
Recommendation system to enhance better customer experience of online shopping.
 TABLE OF CONTENTS

Chapter Particulars Page No.

1. INTRODUCTION 1-20

2. PROJECT METHODOLOGY 20-28

3. THE SITUATION ANALYSIS 30-61

4. FINDINGS AND CONCLUSION 62-64

5 REFERENCES 65
CHAPTER 1
 INTRODUCTION

Industry Overview

About E-commerce:

E-commerce is a type of industry where the buying and selling of products or services are
conducted over electronic systems such as the Internet and other computer networks.
This is an effective and efficient way of communicating within an organization and one of the
most effective and useful ways of conducting business.
It is a Market entry strategy where the company may or may not have a physical presence.
The information technology (IT) sector is a rapidly evolving field that is transforming the
shape of all business standards. The IT sector includes the development of software,
consultancies, software management, online services, and business processes outsourcing
(BPO).

The major different types of E-Commerce are:

I. Business-to-Business (B2B).

II. Business-to-Consumer (B2C).

III. Business-to-Government (B2G).

IV. Consumer-to-Consumer (C2C).

V. Mobile Commerce (M-Commerce).

About Recommendation system:

A Recommender System refers to a system that can predict the future preference of a set of
items for a user and recommend the top items. One key reason why we need a recommender
system in modern society is that people have too many options to use due to the prevalence of
the Internet. In the past, people used to shop in a physical store, in which the items available
are limited. For instance, the number of movies that can be placed in a Blockbuster store
depends on the size of that store. By contrast, nowadays, the Internet allows people to access
abundant resources online. Netflix, for example, has an enormous collection of movies.
Although the amount of available information increased, a new problem arose as people had
a hard time selecting the items they wanted to see.

Recommendation engine:
Recommendation engines have been an important part of the information and e-commerce
ecosystem. They represent a powerful method for enabling users to filter through large
information and product spaces.
Close to two decades of research on collaborative filtering have led to a varied set of
algorithms and a rich collection of tools for evaluating their performance. The differing
personalities exhibited by different recommender algorithms show that recommendation is
not a one-size-fits-all problem. Specific tasks, information needs, and item domains
represent unique problems for recommenders, and design and evaluation of recommenders
needs to be done based on the user tasks to be supported.
Effective deployments must begin with careful analysis of prospective users and their goals.
Based on this analysis, system designers have a host of options for the choice of algorithm
and for its embedding in the surrounding user experience

The recommendation system methods.

Recommender algorithms:

7
The downside of content filtering is that it needs the side information for each item. For
example, genres such as romance and action are the side information of movies. It is very
expensive to have someone watch movies and add side information for every movie out there
and every movie in the future.

The recommendation Techniques available are :

1.Collaborative filtering recommender systems:

Collaborative filtering (CF) is a popular recommendation algorithm that bases its

predictions and recommendations on the ratings or behavior of other users in the system. The
fundamental assumption behind this method is that other users’ opinions can be selected and
aggregated in such a way as to provide a reasonable prediction of the active user’s preference.
Intuitively, they assume that, if users agree about the quality or relevance of some items, then
they will likely agree about other items — if a group of users likes the same things as John
Smith, then John Smith is likely to like the things they like which he hasn’t yet seen.
The information domain for a collaborative filtering system consists of users which have
expressed preferences for various items. A preference expressed by a user for an item is
called a rating and is frequently represented as a (User, Item, Rating) triple.
These ratings can take many forms, depending on the system in question. Some systems use
real- or integer-valued rating scales such as 0–5 stars, while others use binary or ternary
(like/dislike) scales.
Unary ratings, such as “has purchased”, are particularly common in e-commerce deployments
as they express well the user’s purchasing history absent ratings data. When discussing unary
ratings, “purchased” is used to mean that an item is in the user’s history, even for non-
commerce settings such as web page views.

One of the most popular algorithms is collaborative filtering, a class of methods that
recommend items to users based on the preferences other users have expressed for those
items. E.g. Amazon in the US, Flipkart in India etc.

Collaborative filtering (CF) based recommender systems must interact with the user, both to
learn the user’s preferences and provide recommendations; these concerns pose challenges
for user interface and interaction design. Systems must have accurate data from which to

8
compute their recommendations and preferences, leading to work on how to collect reliable
data and reduce the noise in user preference data sets.
Based on purchase history, browsing history, and the item a user is currently viewing,
Amazon recommends items for the user to consider purchasing. Various companies are
finding this as a significant motivation to be able to increase sales volume — customers may
purchase an item if it is suggested to them but might not seek it out otherwise.

Collaborative filtering architecture

2.Content-based filtering (CBF) based recommender systems is another class of systems

which uses content-based approaches based on information retrieval, Bayesian inference, and
case-based reasoning methods. These methods consider the actual content or attributes of the
items to be recommended instead of or in addition to user rating patterns. CBF systems
perform recommendations, such as finding items similar to the items liked by a user using
textual similarity in metadata.
Basically, The goal of a recommendation system is to predict the scores for unrated items of
the users. The basic idea behind content filtering is that each item has some features x. For
example, the movie “Love at last” is a romance movie and has a high score for feature, but a
low score for feature.

9
Content-Based filtering and collaborative filtering

3.Hybrid recommender systems are nowadays more standard. They use collaborative as
well as content based recommendations and have emerged frontrunner as various
recommender strategies have matured, combining multiple algorithms into composite
systems that ideally build on the strengths of their component algorithms.
Collaborative filtering, however, has remained an effective approach, both alone and
hybridized with content-based approaches.

Recommender systems for E-commerce:

The rapid growth of emerging technologies and the Internet has moved the world towards an
e-world where most of the things are digitized and available on a mouse click. Most of the
commercial transactions are performed on the Internet with the help of on-line shopping. The
huge amount of data puts an extra overload to the user in performing on-line tasks.
Recommender Systems are being used widely to reduce this extra overload and recommend
the scrutinized product to the customers. Several data mining techniques are frequently being
used for recommendation technology to enhance the online business, amongst which
Collaborative filtering, Association rules and web mining are on top . In this paper we try to
give an overview of these recommendation techniques with suitable examples and illustrative
diagrams and change of trends in recommender systems. Also the SWOT analysis is

10
discussed for these technologies that give an idea about the respective effects of these
systems on business strategies. Various diagrammatic representations are illustrated. And
finally we conclude that there is a need for an extra effort to overcome the limitations in
existing technology. Also the aspects of these technologies to be used as a business strategy
are going to be discussed further in the report.

About Indian Commerce:

E-commerce has transformed the way business is done in India. The Indian E-commerce
market is expected to grow to US$ 200 billion by 2026 from US$ 38.5 billion as of 2017.
Much of the growth for the industry has been triggered by an increase in internet and
smartphone penetration. The ongoing digital transformation in the country is expected to
increase India’s total internet user base to 829 million by 2021 from 636.73 million in FY19.
India’s internet economy is expected to double from US$ 125 billion as of April 2017 to US$
250 billion by 2020, backed primarily by E-commerce. India’s E-commerce revenue is
expected to jump from US$ 39 billion in 2017 to US$ 120 billion in 2020, growing at an
annual rate of 51%, the highest in the world.

Enhanced by rising smartphone penetration, the launch of the 4G network, and increasing
consumer wealth, the Indian E-commerce market is expected to grow to US$ 200 billion by
2026 from US$ 38.5 billion in 2017. Online retail sales in India are expected to grow 31% to
touch US$ 32.70 billion in 2018, led by Flipkart, Amazon India, and Paytm Mall.

Smartphone shipments in India increased eight% y-o-y to reach 152.5 million units in 2019,
thereby making it the fastest among the top 20 smartphone markets in the world.

Internet penetration in India grew from just 4% in 2007 to 52.08% in 2019, registering a
CAGR of 24% between 2007 and 2019. The number of internet users in India is expected to
increase from 687.62 million as of September 2019 to 829 million by 2021.

Key players in the E-Commerce industry

• Flipkart

• Snap deal

11
 • Myntra

• BigBasket

As per the above players in the market, Since Myntra is a parent organization of flipkart and
has been taken over by flipkart in 2014 in a deal valued at ₹2,000 crore (US$280 million).
The purchase was influenced by two large common shareholders Tiger Global and Accel
Partners. Myntra functions and operates independently.Flipkart-Myntra merger deal has
significant bearing on the e-commerce landscape and startup ecosystem. With the acquisition,
Flipkart becomes the largest e-commerce player by a fair margin in the country and gives it
breathing space in a crowded market. In addition to that In July, 2016 Flipkart acquired
Jabong through its unit Myntra for about $70 million. In February, 2020, Flipkart formally
shut down Jabong to shift focus completely on its premium clothing platform Myntra.So in
general words, Myntra is the parent organization of Flipkart. Talking of Snapdeal, As of
2018, Snapdeal had a market share of 1.9 percent, which is no longer in the competition of
Flipkart. Currently, The major competitor for Flipkart is Amazon India which is actually
called as Amazon.com, Inc., is an American multinational technology company based in
Seattle, Washington, which focuses on e-commerce, cloud computing, digital streaming, and
artificial intelligence.In 2018, Amazon was the second-largest online retailer in India,
trailing Flipkart, with 32 percent market share (compared with 41 percent in the United
States). And analysts at Forrester predict e-commerce sales in the country will reach nearly
$86 billion by 2024.

COMPANY OVERVIEW

About Flipkart
Flipkart is the leader of e-commerce in the fast-growing Indian market founded in 2007 by
Sachin Bansal and Binny Bansal. The company focuses on goods such as electronics, books,
music, and movies. Consumers can access the site online or through an app on their mobile
phones, before expanding into other product categories such as consumer electronics, fashion,
and lifestyle products. As the Leading E-commerce giant of India, Flipkart has 75 million
registered users who have helped the company to achieve 5 billion-dollar GMV (Gross

12
merchandise value) of sales in FY15. GMV is the indicator of the performance of the
company in terms of the total value of merchandise sold by the company during the period. In
recent years, Flipkart has increased its budget spend on digital marketing considerably.
Earlier, they were spending crores in multichannel marketing campaigns that resulted in a
large amount of cash burn. Since Kalyan Krishnamurthy took over Flipkart in 2017, there has
been a large-scale revamp of its marketing strategy - it has become leaner and heavily based
on digital, which has worked out well for them.

Flipkart Products and Services Offered

● Flipkart Plus Membership

A points-based service on Flipkart is Flipkart plus Membership that lets its users earn
‘Plus Coins ‘with every purchase. Flipkart Plus intends to reach customers with a
range of options on rewards, and a no-fee membership. To get Flipkart Plus
Membership, the users should have digital Coins.

● Flipkart Student Membership

Flipkart offers a one-year free Student membership. To avail of the student

membership offer on Flipkart, you just need to upload a valid college ID of yours.

● Flipkart Coins

Flipkart coins are the digital points that you can earn for every purchase equal or
greater to 250. The maximum coins you can earn for a single purchase is 10 coins.

● Flipkart Super Coins

Flipkart Super Coins is the digital rewards ecosystem that is designed to provide
benefits to Flipkart customers.

● Flipkart Gift Card

Quicksilver Solutions Pvt. Ltd. issued a Flipkart gift card, which is also known as a gift
voucher. It is a prepaid stored-value money card, which can be used as an alternative
to cash for purchases on Flipkart.

13
 ● Flipkart Credit Card
You will get instant cashback with every transaction by using Flipkart Axis Bank
Credit Card.

● Flipkart Assured

flip kart Assured is a new quality and speed badge that is applied to select products
in key categories. It is the new benchmark in online shopping which guarantees that
high-quality products are delivered safely to its customers in the possible fastest
time.

● Flipkart Pay Later

Flipkart Pay Later is another familiar service of Flipkart that is extended by select
sellers to select customers. The sellers allow their customers to pay later for their
orders.

About Myntra

Myntra is an Indian fashion e-commerce company headquartered in Bengaluru, Karnataka,

India. The company was founded in 2007 to sell personalized gift itemsMyntra.com is a one-
stop-shop for all your fashion and lifestyle needs. Being India's largest e-commerce store for
fashion and lifestyle products, Myntra.com aims at providing a hassle-free and enjoyable
shopping experience to shoppers across the country with the widest range of brands and
products on its portal. Myntra happened to be India's first online portal for personalized
merchandise and was the most successful brand in the aforesaid segment. The firm which
started with personalized merchandise solutions to companies realized that the business was
fragile with wafer-thin margins and large volumes. The CEO studied the Indian market and
found that the fashion and lifestyle industry is a $30billion market in India and decided to
become an online apparel merchandiser. In 2010, the company shifted its strategy to
becoming B2C (Business to Consumer) and re-invented itself into myntra.com expanding its
catalog to fashion and lifestyle products.

14
Myntra Products and Services Offered

Myntra.com is India's largest online fashion store; it is one of the first of its kind in India as
most of the online fashion and lifestyle store is mere online shops where you just pick and
pay whereas in Myntra.com customers are provided with:

A rich detailed catalog of products having clear creative's and clear call of action with 24
hours of customer support services.

The website elements i.e. menu layout, access to product information, logical presentation,
screen design, navigation, and presentation of product attributes and product availability are
well maintained ensuring a comfortable user interface.

Myntra retails only the most popular in-season merchandise.

Myntra offers a 30-day return policy which gives an ample amount of time to the Indian
customers to touch and feel the product

Myntra attempts to ship every order within 24 hours, free shipping within India on all
products. It can also ship goods internationally to all major countries

Payment terms would be COD (Cash on delivery) and Credit/debit cards

About Snapdeal

Snapdeal is one of India’s leading e-commerce companies with its headquarters located in
New Delhi. Snapdeal was launched in 2010, a time when the e-commerce market in India
was at a nascent stage. As a daily deal platform but expanded in September 2011 to become
an online marketplace. Snapdeal has grown to become one of the largest online marketplaces

15
in India. In March 2015, Snapdeal brought actor Aamir Khan for the promotion of its website
in India.
The company was co-founded by Kunal Bahl and Rohit Bansal. Snapdeal currently offers
more than 60 million products across various categories such as mobiles & tablets,
computers, office & gaming, electronics, home & living, men’s and women’s fashion, sports,
fitness & outdoors, daily needs, motors & accessories, books, music, real estate, and financial
services. The company has more than 3 lakh sellers on its e-commerce platform that cater to
millions of users. Snapdeal has a wide logistics network and it delivers to more than 6000
cities and towns in India.

Snapdeal is one of the most funded companies in India. To date, it has received total funding
worth $1.78 billion. Snapdeal has received funding from individuals as well as private equity
investors and venture capitalists. Some of the top investors in Snapdeal include SoftBank,
BlackRock, Temasek, Foxconn, Alibaba, eBay Inc., Premji Invest, Intel Capital, Bessemer
Venture Partners, Mr. Ratan Tata, Clouse SA, Ontario Teachers' Pension Plan, Kalaari
Capital, Nexus Venture Partners, Cambrian Ventures, Iron Pillar, and Myriad Group.

Snapdeal Products and Services Offered

Snapdeal has gained a lucrative platform, which it utilizes to market at least five
million products in a range line that is as diverse as the country. It is the second-largest
company related to e-marketing with more than fifty thousand merchants attached and
working through it. The company has created a buzz by using this platform to feature more
than six thousand brands in five hundred categories. "Xiaomi”, the famous brand has cracked
a deal with Snapdeal under which it will sell its product at this site. Some of the most liked
and purchased products and their categories are as follows-

Apparel – this includes clothing for men, children, teenagers, and women from brands like
Biba, Rangriti, etc.

Personal care and Beauty products – brands like Maybelline, Lakme, Livon, and L’Oreal are
parts of Snapdeal with products related to oral care, make-up, skincare, haircare, and
grooming.

Fashion Accessories

Perfumes and Beauty – This includes deodorants Attars, air fresheners, and perfumes for
men and women.

16
Health & Nutrition- This includes medical equipment, massagers, etc.

Mobiles & Tablets–This includes Memory cards, earphones, tablet accessories, Bluetooth
devices, power banks, and tablets, phones belonging to brands like Apple, Micromax, and
Samsung

Footwear – This includes shoes for men, children, women from important brands like Puma,
Sketchers.

Jewelry – This includes fashion jewelry, gold coins, and silver jewelry, etc.

Computers & Laptops – This includes monitors, desktops printers, etc.

Music Instruments – This category includes guitars, drums, DJ equipment, etc.

Books- This includes audiobooks, e-books.

Movies

Camera and its Accessories

Home & Kitchen – brands like Pigeon, Tummerware, and prestige display their products
like cookware, disposables, glassware, and catering supplies.

Big Basket: India’s Largest Online Supermarket

Bigbasket.com (Innovative Retail Concepts Private Limited) is India's largest online food and
grocery store. With over 18,000 products and over 1000 brands in our catalog, you will find
everything you are looking for. Right from fresh Fruits and Vegetables, Rice and Dals,
Spices, and Seasonings to Packaged products, Beverages, Personal care products, Meat – we
have it all.Choose from a wide range of options in every category, exclusively handpicked to
help you find the best quality available at the lowest prices. Select a time slot for delivery and
your order will be delivered right to your doorstep, anywhere in Bangalore, Hyderabad,
Mumbai, Pune, Chennai, Delhi, Noida, Mysore, Coimbatore, Vijayawada-Guntur, Kolkata,
Ahmedabad-Gandhinagar, Lucknow-Kanpur, Gurgaon, Vadodara, Visakhapatnam, Surat,
Nagpur, Patna, Indore, and Chandigarh city. Where customer can pay online using your
debit/credit card or by cash / Sodexo on delivery.We guarantee on-time delivery and the best
quality

17
About Big Basket:

bigbasket.com allows customers to walk away from the drudgery of grocery shopping and
welcome an easy relaxed way of browsing and shopping for groceries. Discover new
products and shop for all your food and grocery needs from the comfort of your home or
office. No more getting stuck in traffic jams, paying for parking, standing in long queues, and
carrying heavy bags – get everything you need, when you need, right at your doorstep. Food
shopping online is now as easy as every product on your monthly shopping list is now
available online at bigbasket.com, India's best online grocery store.

Big Basket Products and Services Offered

Big Basket is a proud seller of more than 18,000 products delivered every single day. The
brand houses products coming from 1,000 different sellers and provides them with a platform
to showcase and sell their products. Big Basket eliminates the role of go-betweens in the
consumer-supplier chain. It saves a lot from the money of the customers.

The products manufactured by some of the best brands can be bought just within a few clicks.
It has worked in favor of the working people who find hardly any time to go grocery
shopping.

CHAPTER 2
PROJECT METHODOLOGY

The objective of the Study

18
The study will focus on the Recommendations system employed by the two big Indian
originated e-commerce companies- Flipkart/Myntra and Bigbasket. The study will make a
thorough comparison of the filtering techniques,remarketing, recommendation engine for
marketing and advertising, predictive analysis, customer engagement through R s and various
qualitative aspects of the selected companies. There will be a focus on the strategies and
algorithms/Filtering techniques followed by each organization over years, and how they have
been able to make a mark for themselves riding on their respective strengths through the
recommendation system.

Exploratory Research methods:

Several methods of exploratory science are essential for data collection and analysis.
Exploratory work was, however, divided into two main types, namely the primary and
secondary types of study. The method of doing research appears to be more complicated
when dealing with a problem that wasn't studied in detail before. Data is obtained directly
from the topic of inquiry in the primary research methods. In this case the theme, maybe a
group of people, or an individual.

Secondary Research methods

The methodology followed will be a Secondary analytical approach that makes use of the
current methods on the subject under analysis. Existing outlets such as newspapers, journals,
posts, papers etc are what researchers conduct for exploratory work. The details used for
publications must be quoted. Some of the secondary methods to forensics used in exploratory
study. In the secondary research, I have gone through the best recent article/papers on the
particular topic.

Literature

Literature research is the way old instruments, such as journals, textbooks, letters, magazines,
etc., are managed. Each of this information will be collected in both sift copy and hard copy
documents.

Online Sources

19
This research has gained much prominence among millennials, with the advent of
technology. The cheapest and simplest form of study is by online data outlets.

A researcher can navigate through as many resources as possible, with internet access and a
personal computer or cell phone. They can also be downloaded in future for further use. The
drawback of this approach is the challenge of combining truthful knowledge across the many
online services. Researchers face the risk of ending up with inaccurate data, as it may be
difficult to detect false information.

Data Collection

The data collected for the preparation of the project report includes secondary data.
Secondary data is collected through annual reports, text collected from the website. It
includes research papers for the past few years.

Research Objectives

● The Research presents an overview of recommender systems that covers the framework and
approaches and associated issues and problems by Indian E-commerce companies.
● Effect on revenue of the Indian e-commerce companies due to the recommendation system.

Issues raised in research

● How recommendation systems work on Indian E commerce companies.

● Recommendation systems in different industry sectors

● Types of recommendation systems used by Indian E commerce companies

● Recommendation systems and Ecommerce

20
21
 LITERATURE REVIEW

E-Commerce websites are the major emerging trends in the current scenario, which facilitates online
product selection, purchase, and sales. Nowadays E-commerce websites have better popularity and
advent nature, so numerous users wish to share their opinions about their experience in the form of
making reviews, ratings, and blogs.

In the last twelve years, the number of web users increases, so intensely leading to intense
advancement in web services which leads to enlargement of the usage data at higher rates. The
purpose of a Recommender System is to generate meaningful recommendations to a collection of
users for items or products that might interest them. Recommender systems differ in the way they
analyze these data sources to develop notions of congeniality between users and items which can be
used to identify well-matched pairs. The recommender system technology intentions to help users in
finding items that match their interests. It has a successful usage in e-commerce applications to deal
with problems related to information overload proficiently. In this paper, we will extensively present a
survey of six existing recommendation systems. The Collaborative Filtering systems analyze
historical interactions alone, while Content-Based Filtering systems are based on profile attributes,
Hybrid Techniques attempt to combine both of these designs, Demographic Based Recommender
systems aim to categorize the user based on personal attributes and make recommendations based on
demographic classes, while Knowledge-Based Recommendation attempts to suggest objects based on
inferences about a user's needs and preferences, and Utility-Based Recommender systems make
recommendations based on the computation of the utility of each item for the user. Finally, few
research papers had an influence on research-paper recommender systems in practice. We also
recognized a lack of authority and long-term research interest in the field, 78% of the authors
published no more than one paper on research-paper recommender systems, and there was miniature
cooperation among different co-author groups.

The basic concept behind the “recommendation system “aims to provide users with personalized
online product or service recommendations to handle the increasing online information overload
problem and improve customer relationship management. Various recommender system techniques
have been proposed since the mid-1990s, and many sorts of recommender system software have been
developed recently for a variety of applications. Researchers and managers recognize that
recommender systems offer great opportunities and challenges for business, government, education,
and other domains, with more recent successful developments of recommender systems for real-world
applications becoming apparent.

22
It is thus vital that a high-quality, instructive review of current trends should be conducted, not only of
the theoretical research results but more importantly of the practical developments in recommender
systems. This paper, therefore, reviews up-to-date application developments of recommender systems,
clusters their applications into eight main categories: e-government, e-business, e-commerce/e-
shopping, e-library, e-learning, e-tourism, e-resource services, and e-group activities, and summarizes
the related recommendation techniques used in each category. It systematically examines the reported
recommender systems through four dimensions: recommendation methods (such as CF),
recommender systems software (such as BizSeeker), real-world application domains (such as e-
business), and application platforms (such as mobile-based platforms). Some significant new topics
are identified and listed as new directions. By providing state-of-the-art knowledge, this survey will
directly support researchers and practicing professionals in their understanding of developments in
recommender system applications.

A lot of Recommender System (RS) have followed the above-mentioned factors for the finest product
suggestion to the users. Although the results are best and reliable, the e-commerce system should take
extra considerations on the related/similar product analysis. The personalization cannot be determined
with only product similarity, this also needs to be identified by their personalized features and
interest. So, the proposed system performs effective product recommendations and increases customer
satisfaction.

"Onto Commerce: an ontology focused semantic framework for a personalized product

recommendation for user-targeted e-commerce” (29 March 2019) Gerard Deepak,
Dheera Khasaraneni

This research is regarding the paradigm shift in business strategy in terms of online marketing and e-
commerce and to comprehend the World Wide Web transforming into an intelligent semantic web,
there arises a perpetual need for semantically driven e-commerce systems which gives preference to
the users. In this paper, Onto Commerce which is an e-commerce system that incorporates semantic
algorithms for product recommendation has been proposed. The proposed strategy uses the enriched
normalized pointwise mutual information measure for semantic similarity computation. Onto
Commerce assimilates ontologies and recommends products based on the user query, recorded user
navigation as well as user profile analysis thereby encompassing personalization. To make the
recommendations more relevant, Onto Commerce uses parametric fuzzification to increase the
number of relevant recommendable products. Onto Commerce yields an average accuracy of 88.68 %
with a low false discovery rate of 0.13 which makes it a best-in-class semantically driven product
recommendation system for online e-commerce.

23
“Collaborative Filtering Techniques in Recommendation Systems” Sandeep k Raghuvanshi,
R.K. Pateriya (19 March 2019) in his paper, analyzed that the Recommender system is the tool
to user preferences over a given set of items. It takes the help of the previous auxiliary
information in terms of feedback or ratings. The main purpose of a recommender system is to
engage users and enhance their experience over the Internet. Presently, recommender systems
are widely used over e-commerce and social networking sites. The different applications
require specialized recommendation system for them as e-commerce sites recommendation
systems are different from social networking sites. So, the recommendation system's biggest
challenge is diversity as one cannot generate an accurate prediction using the same technique
for different applications. This paper is an effort to illustrate one of the popular
recommendation techniques, collaborative filtering based on classes, memory-based, and
model based on two popular data sets (Movie lens and Jester). Further, it represents a
comparative analysis of how results diverge from application to application and provide a
way to optimize the results of existing algorithms to get the most out of them. The purpose is
to present an exposure and open door to use more sophisticated data mining and machine
learning techniques to enhance the overall efficiency of the recommendation system.

Daniel Kraus, Victoria Reibenspiess, Andreas Eckhardt (August 01, 2019) in their paper “How
Voice can change customer satisfaction: A comparative analysis between E-commerce and
Voice Commerce
Found that Voice commerce is a newly evolving e-commerce channel where consumers communicate
with dedicated systems on smart speakers or other devices using their voice, to find products. This
paper comparatively investigates factors for customers' satisfaction in voice commerce and e-
commerce. Being the first study to scientifically analyze customer satisfaction factors in voice
commerce and compare them with e-commerce, we surveyed 178 consumers and used structural
equation modeling for statistical hypotheses testing. The results show that consumers have higher
expectations of inconvenience for voice commerce than they have for e-commerce. Transaction
process efficiency significantly influences satisfaction in voice commerce, but not in e-commerce.
This research provides implications for future research on voice commerce strategy and system
design.

Anitha Anandhan, Liyana Shuib, Maizatul Akmar Ismail, Ghulam Mujtaba in their paper “Social
Media Recommender Systems: Review and Open Research Issues” (27 February 2018)
This paper found and analyze that different types of review systems have been developed

24
with the recommender system (RS). RS is developed based on user textual reviews, ratings,
and comparative opinions. RS for social media resources, such as blogs, forums, social
network websites, social bookmarking websites, video portals, and chat portals help users to
collaborate effectively. Social media resources are used in the RS for recommending content,
articles, news, e-commerce products, and users. Although research on social media in RS has
increased annually, comprehensive literature review and classification of these RS studies are
limited and must, therefore, be improved. In the results, they conclude that present results can
contribute significantly to understanding social media in RS. A total of 61 articles out of 434
were rigorously selected and reviewed. The classification of social media RS research in
terms of recommendation approach, domain, data mining technique, dataset, recommendation
type, and performance metric can also provide detailed insights into these aspects and
guidelines for future RS research in social media.

Saumya Bansal, Niyati Baliyan " A Study of Recent Recommender System Techniques"
(2019) analyzes that The influx of data in most domains is huge and dynamic, leading to big
data and hence the need to build a recommender system grows stronger. This work is a
comprehensive survey of the status of different recommendation approaches, their
limitations, and extension which when applied may eradicate the incessant information
overload problem of the web. Further, an investigation is conducted on the Google Scholar
database, delineating the temporal distribution of different recommendation techniques.
Several popular and most-used evaluation metrics, domain-specific applications, and data
sets used in the recommendation are reviewed. By summarizing the current state-of-the-
art, this work may help researchers in the field of recommendation system techniques and
provides future directions highlighting issues that need to be

25
 CHAPTER 3
1. Situation Analysis

Role of Recommendation System in E-commerce companies:

Why the Recommendation system?

Recommendation systems are one of the primary ways in which any e-commerce websites
tend to generate repeat purchases, that is, getting a purchase from an already registered
customer. Repeat purchase is one of the key metrics that e-commerce websites measure since
a repeat customer means less money spent on marketing to get him/her to make a purchase.
Recommendation systems are not a new technology. These have been around for as long as
there have been online shopping. The most talked about the recommended system is that of
Amazon. The technology and its working are visible through sections like Recommended For
You, Frequently Bought Together, or Based On Your Previous Purchases Following Is
Recommended For You, among others.
Although the system has been around for a while, it seems to have not evolved with time. By
evolution, I mean changes in recommendation systems have been largely pushed back. With
this article, I wish to discuss what can be done on current recommendation systems within the
framework of already existing algorithms.

The purpose to use the Recommendation System

● Assistance in decision making

● Assistance in comparison
● Assistance in discovery
● Assistance in exploration

Working of Recommendation system:

What is the Recommendation system?

Recommendation systems have different algorithms. A simple recommendation system looks
for users who have made the same purchases as you and rated these items similarly. The
system then removes already-purchased items from the list of other items these similar
customers purchased and rated highly. The rest of the items are then recommended to you.
Two popular versions of these algorithms are collaborative filtering and cluster models.

\
Phases of recommendation system

43
Understanding the Relationships

Relationships provide recommender systems with tremendous insight, as well as an

understanding of customers. Three main types occur:

User-Product Relationship

The user-product relationship occurs when some users have an affinity or preference towards
specific products that they need. For example, a cricket player might prefer cricket-related
items, thus the e-commerce website will build a user-product relation of player->cricket.

Product-Product Relationship

Product-product relationships occur when items are similar, either by appearance or

description. Some examples include books or music of the same genre, dishes from the same
cuisine or news articles from a particular event.

User-User Relationship

44
User-user relationships occur when some customers have similar tastes concerning a
particular product or service. Examples include mutual friends, similar backgrounds, similar
age, etc.

Recommendation system generating revenue through Product cross-sell and up-sell:

Since e-commerce companies have plunged into the ever-increasing competition, companies are
always on the lookout for ways to increase their sales figures. Since businesses typically have a large
number of products for sale, therefore the use of recommender systems in an e-commerce
environment improves financial performance by helping customers find the right product.

Recommender systems are reshaping the business world, especially E-commerce, in an unprecedented
manner. They learn patterns of behaviors to predict someone's preference on a set of items they have
not experienced. Over the past few years, thousands of companies have used recommender systems to
help their customers find products to purchase, new movies to watch, songs to listen to, and even
people they should interact with.

The application of recommender systems that we will focus on is product up-sell/cross-sell.

The goal is to suggest upgraded and/or additional products for customers to purchase. If our
recommendations are successful, the average order size per customer should increase, thus increasing
revenue.
For example:
when a customer purchases a new iPod. A common cross-sell would be to suggest also buying new
headphones

● Cross-selling through RS:

Recommender systems improve cross-selling by suggesting additional items
associated with the products in the shopping cart before the checkout process. This
brings the customer back to the shelves and thus increases the average order size.
● Up-selling through RS:
RS can improve Upselling by suggesting a superior model for a product under
consideration, In other words, Up-selling means recommending a similar but more
expensive product to the customer.
For example, if a customer is looking for an 8 GB pen drive, offer him 16 GB at a
discount.

45
A study of Recommendation system employed by Flipkart, Snapdeal,
Myntra and Big Basket: The Key Players in Indian E-commerce Industry

Recommendation system and Indian E-commerce:

The E-commerce business in India has seen exponential growth over the last decade. This
growth is due to many contributory factors, including the rapid adoption of technology by
Indian consumers, large increases in the number of internet users, new enabling technologies,
innovative business models, and alternative payment options offered by E-commerce
companies. Moreover, the high growth in E-commerce continues unabated, with the sector
expecting to witness a steep increase in revenues in the coming years

● India is one of the fastest-growing eCommerce markets worldwide with millions of

new internet users taking advantage of cheap mobile connections to send text
messages, stream online content, and shop online.
● As of 2020, 25% of the local population was using the internet; almost ten times the
user base from a decade prior.
● Mobile phone internet user penetration in India is projected to reach 36% of the
population in 2021, representing a huge potential in terms of digital and mobile buyer
audience.
● With a population of 1.3 billion, India currently has an estimated 490 million mobile
internet users in 2018. This user base is estimated to surpass 829 million in 2021.

Significance of Recommendation system in Indian E-commerce:

If any of Indian E-commerce companies want client interaction and sales on your e-
commerce website, then, a recommender system is what they need and These intelligent
systems use deep learning technology to make accurate product recommendations and can be
easily integrated into the user’s existing website or mobile app.

The following are the significance of an e-commerce recommendation engine:

● Better User Experience:

Collaborative filtering - creates recommendations by comparing the current user to other
users that have similar preferences.

46
 Content-based filtering - makes recommendations based only on the specific user's data,
such as their past purchases or searches.
● Sales Boost: Through optimizing the website will ensure boost the sales by increasing
the conversion rates and promoting the long-tail inventory. The customers will keep
returning to the site because when you know what they want, their shopping journey
becomes easier.
● Enhanced Customer Engagement: Improved customer experience will guarantee a
higher customer engagement rate. When the customer feels valued and, at the same
time, finds your website user-friendly, your chances of engagement will be enhanced.
Keeping a one-to-one relationship with every customer is simply impossible,
however, that does not mean you cannot engage them in a meaningful way.

● More traffic: The improved customer experience, better engagement, as well as the
strategies that boost the sales rate create more internal traffic on your website. The
greater are the performance numbers, the more is your internal traffic. People who
click on recommendations visit more pages, putting a higher number of products into
the carts.

Recommendation system employed by Flipkart:

Why Flipkart?

Since Flipkart is the national leader of e-commerce in India. Founded in 2007, Flipkart was initially
selling books, before it expanded to other popular categories including mobile phones, electronics,
fashion, and lifestyle products. Flipkart also held a 39.5% market share of India's e-commerce
industry.

As Flipkart has a huge contribution to the Indian market, The Company had its constant and as well
as frequent customer base who spent more time on their website to search the related products
based on their taste.

To grab and sustain their footfalls, The company needed Recommendations who play a pivotal role
in being the 'Online Shopping Assistant' to users via helping them discover and narrow down the
relevant selection. 'Similar products' is one such Recommendation module on Flipkart that helps

47
users discover products similar to the ones they have browsed.

The data analytics and algorithm behind RS at Flipkart:

Scope and Usage:

● The scope of data analytics grows exponentially each day, many e-commerce
companies including Flipkart use them as "Tools" designed to meet customer needs
and taste to optimize the business processes. This is helping the companies to make
better decisions, and the recommender system helps the data analyst of Flipkart to
succeed in it.

● Data analytics also helps to identify both loyal and new customers by using data
extraction and segmentation for tracking browsing habits and spending patterns. The
technology allows e-commerce companies to customize their offerings and
promotions.

● They analyze every click and touch in every user session to construct something
called the 'journey' of each customer and helps them understand their flow through the
Flipkart app and predict their next purchase.

For Example,When a customer of Flipkart researches through the catalogs of a

product range, the search patterns are recorded and a persona is created for the
customer so that when he returns to the website, the searching time is drastically
reduced by showing the most relevant product that the customer could be interested in
buying.

Another example, When a flip kart customer buys a mobile phone, out of the millions
of products in the catalog, the system correctly matches the most probable product
you will purchase next - a mobile case or a screen protector – an example of content-
based filtering. Also, the system can recommend products based on the person's
persona or the persona of a similar customer from the same demography.

48
The Filtering technique used by Flipkart: The recommendation algorithm

● As Flipkart being the 'Online shopping Assistant' to its users by helping them discover
and narrow down the relevant selection.
● Flipkart uses "Similar products” as one such Recommendation module that helps
users to discover products like the ones they have browsed. It has one of the highest
conversion rates and is a significant contributor to the company's units and revenue.
They use a hybrid of content and collaborative filtering techniques to generate
“Similar product recommendation”
The content matching is done over product attributes and images in the catalog; and
the collaborative filtering algorithm is applied over users' browse data (like product
page views, Wishlist, add to cart, etc.) to find the most frequently co-browsed
products for a given product. The ranked list of similar products based on relevancy is
obtained from combining these multiple sources.
● But relevance, by itself, is not the sole lever to help users make a purchase decision.
We wanted to further help the user narrow down the selection by showcasing the
‘right’ products within the relevant set.

The following Signals that incorporate their ranking function are as follows:

● Product Quality:
To check whether the material or seller is good and durable also the feedback insights
that other users already brought are some repetitive and constant quality specific
questions that users ask before buying a product. We take these 'Quality' signals into
account and show products that are of higher quality on top of the Recommended list.
This helps the 'trust factor' that our customers build on our recommendations.

● Performance
The more popular and fast-selling the product is, the better the chances for the user to
make a buying decision. We take past performance and popularity as one of the
signals in our ranking to ensure products with good performance are shown to the
user.

49
 ● Diversity
Typically, in the online shopping world, what you see is what you click; what you
click is what is fed to the system and maybe re-shown to you and other users. We
need to break this vicious circle, we continuously iterate on this front to bring in some
diversity to the products they recommend, to better engage the users.
They have introduced a new layer of ranking mentioned in the attached picture
below, They also use a "Machine learning" model that takes in the above-mentioned
signal or many other signals as features and ranks the products to optimize for
"Engagement and conversion"

In the above picture, the Level 1 Ranking is optimized for relevance and Level 2 Ranking is
optimized for performance. They use Point-wise Logistic Regression in their model that can
predict the click-through rate and conversion of the products and rank them per our
optimization function. The model can operate in batch mode as well as in real-time and uses
over two million features to rank the Recommended products.

50
The Level 1 ranker consumes multiple terabytes of data across all users and products in our
catalog to generate the relevance list. With the data corpus expanding every day, they ensure
that their underlying system performs to keep pace with the emerging data.

As a result of the following technique/algorithm

The following are the outputs:

● They have seen a 10% and the increase in the click rate of 3% in conversion of this
module and the whole model has become much easier to maintain.
● They also noticed a quick addition of any new feature that can influence
recommendations and observed the entire engine automatically learn the performance
of the feature.

Recommendation system employed by Big Basket:

Why Big Basket?

● Since 2011, Bigbasket has been the largest online grocery supermarket in India and
about the time when e-commerce was in its nascent stage in the country

● As one of the Key players in Indian e-commerce for online shopping Big Basket
Business Model covers most of the prominent cities in India.
Some of the most popular cities where Big Basket is functional are Hyderabad, Pune,
Mumbai, Chennai, Delhi, Mysore, Noida, Coimbatore, Kolkata, Vijayawada-Guntur,
Ahmedabad-Gandhinagar, Lucknow-Kanpur, Gurgaon, Visakhapatnam, Vadodara,
Surat, Nagpur, Indore, Patna, and Chandigarh Tricity. Till 2019, Bigbasket had the
highest market share among online groceries in India at over 35 percent.

● As Online grocers are going all out to attract modern shoppers, and supporting them
in this endeavor are new-age technologies. Bigbasket is using artificial intelligence,
machine learning, and also analytics to streamline strategies and improve customer

51
 experience.

● As per the Estimation, About 30% of Bigbasket customers place orders through smart
phones. Unlike other e-commerce companies such as Amazon, Bigbasket customers
place orders for several products in a single order, sometimes as high as 80 in one
order depending on their purchase frequency. When the basket size is high, using
smartphones to place orders is challenging.

● It is a common phenomenon that customers forget to place orders for a few grocery
items which may result either in placing additional orders or customers purchasing
those products from neighborhood stores resulting in a financial loss to online grocery
stores.

The data analytics and algorithm behind RS at Big Basket:

Scope and Usage:

● Since E-grocers have been going all out to engage modern shoppers,
being aided in this mission by new-age technologies. Bigbasket is
making use of Artificial Intelligence, Machine Learning, and analytics to
simplify logistics and boost customer experience.

● Data analytics is playing a very pivotal role in boosting

Bigbasket’s market, A couple of the analytics tools that

52
 Bigbasket leverages are Tableau which is leveraged for
converting complex data to simplified versions and Power BI
which is leveraged for interpreting data as well as for sharing
insights on it.

The following are Role of Analytics in driving business in the company are as follows

● Extracting essential information from the unstructured data:

The Bigbasket team has to deal with a huge section of data that is largely
unstructured. This data is obtained from studying the customer’s online behaviors and
interpreting their transaction records. Hence customer analytics helps the company to
comprehend the correlation between delivery matrix and customer loyalty.

● Enabling personalization through analytics:

The analytics team aims at enhancing and personalizing customer

experience by facilitating them with all kinds of solutions. It also helps in
detecting customers as well as boosting engagements such as
communications as well as offers. The analytics aids in studying customer
behavior so that the company can design its personalization according to it.

● Reducing inventory from the flux through predictive analysis :

Predictive analytics helps to make sure that the firm is neither out of stock nor
overstocked. This approach implies forecasting the demand, that will aid the supply
chain team in their planning abilities such as the drivers, vans, CEEs, and equipment
which are employed for various facilities. It also aids in examining the array of orders
and volumes so that the team of the supply chain can strategize accordingly.

53
The Filtering technique used by BigBasket: The recommendation algorithm

Since, They are in an e-commerce business where data is the new driving fuel. They
leverage data-driven capabilities from several channels.
The Data is collected From the following parameters:
Data is collected from transactions, customer preference, shopping behaviour, etc
to build a variety of algorithms (statistical algorithm, Deep Learning algorithm
and ML algorithm).

All these algorithms are used for different use cases as per the following matters
mentioned above.
● Bigbasket banked on Machine Learning and analytics to claim on-time delivery (99%
or more). The process of delivery now starts with identifying the common routes and
vehicle status for packing multiple sets of orders.

● For a successful on-time delivery, the team gets insights from various data sources
such as best route, current traffic situations, number of vehicles required on which
route etc.

● With insightful data channels, the ML-driven model analyzes real-time situation and
customer requirements. The model helps in managing multi-level track of pickers,
reverse pickers, scheduled deliveries etc.

Problem recognition at BigBasket:

Difficulty in using phones to place order when basket size is large:
Customer perspective about the problem:
When a customer enters into their Bigbasket online account, they need to check a
huge list of items even for a small set of required products for them to select into their
cart. The customers’ set of regular products to purchase is usually quite small when
compared to the complete list of products in the website. It takes a big effort and time

54
of about 20 to 30 minutes to place an order as they are being placed on mobile
devices.

Business perspective about the problem:

Bigbasket is also facing a huge problem on their business side as well. When a
customer forgets a product while going through the long order process, he has to place
an order again. This leads to making multiple trips for the same customer’s location
within a span of a few hours thus increasing the cost of logistics for the company.
Sometimes customers choose to buy that forgotten product from a nearby store thus
losing some part of business. This impacts the creation of a smaller basket size in the
future for Bigbasket.com from the customer.

The Filtering technique appropriate to the particular issue at BigBasket:

Since companies like Bigbasket.com face different problems when compared to these e-
commerce companies.They analyzed and found that they can consider the “Content-based
Recommendation” technique. Bigbasket.Com has access to all the sales data through their
customer analytics

Why Content-based Recommendation technique?

As mentioned above, The Big Basket uses Customer Data analytics through their analytical
engine in driving their business. The following are the reasons below

● They have a list of registered customers along with their purchase history. This is the
input to create a recommendation algorithm.
● The content-based approach needs high knowledge of huge inventory of Bigbasket
items and Bigbasket has that huge data suitable for building a recommendation
algorithm. Each item must be characterized and catalogued based on its
characteristics.
● Also The Big basket has very large inventory as they have daily based products like
vegetables, bakery and dairy products and branded foods (this is the only kind of
inventory suits for Content-based Recommendation).

55
The Architecture of RS while ordering by customer

As a result of the following technique/algorithm after implementing at Big Basket

The following are the outputs:

Customers usually forget to buy items while giving an order for more number of items. So,
the idea for the ‘‘Did you forget?” feature has emerged, providing product recommendation
to the customer while checking-out, based on the purchase history of a particular customer.
The “Did you forget?” The feature solves the unique problem that Bigbasket has.

This feature can be created by considering the following approach.

● The Customer’s Purchase history items and the items chosen in the basket is the
data considered to derive the recommendation. This recommendation algorithm
does not concentrate on purchase history or order patterns for other similar
customer profiles.

56
 ● This feature helps the specific situation where the customer has already filled the
basket and is ready to check-out. The preferences of the users are then concluded
based on their searches, behavior (purchase history etc), or directly entering
information about their preferences in the filters.
● In this Content-based approach, the user needs to input information on their
personal tastes, though on the same scale. The good side of the content-based
approach is that it doesn't suffer from the cold-start problem that even the first
user can get the useful recommendations if the products are categorized and
cataloged perfectly.
● The benefit that recommendations offer to the user must justify the effort required
to offer input on personal tastes. That is, the recommendations must be excellent
and the effort required to enter personal preferences must be minimal and ideally
baked into the general usage. (Note that if your offering is an e-commerce store,
this data entry amounts to adding a step to your funnel and could hurt sales more
than it helps.)
Use of Recommendation engine at Big Basket:

They have also created a recommendation engine made using a bunch of analytical streams
like FP development, affinity analysis development, and then collaborative filtering. The
recommendation algorithm helps consumers to check out new items at the same time as
keeping their personal preferences while offering suggestions

Challenges that anticipate while building a recommendation engine for Bigbasket:

 The implementation and deployment challenges of a recommendation engine for Bigbasket are
that an automated classification system could require significant algorithmic development and is
likely not available as a commodity technical solution. The user needs to input information on
their personal tastes, though not on the same scale. 

● Lack of Data

The biggest issue facing recommender systems is that they need a lot of data to
effectively make recommendations. It’s no coincidence that the companies most
identified with having excellent recommendations are those with a lot of
consumer user data: Amazon, Netflix. A good recommender system firstly needs
item data (from a catalog or other form), then it must capture and analyze user
data (behavioral events), and then the magic algorithm does its work. The more

57
 item and user data a recommender system has to work with, the stronger the
chances of getting good recommendations. But it can be a chicken and egg
problem – to get good recommendations, you need a lot of users, so you can get a
lot of data for the recommendations.

● Lack of depth in data

This approach requires a deep understanding of the content itself in the data. The
data of all the products should be classified, categorized and cataloged well. Each
and every attribute of the product has to be captured and well saved in the
database which is the essential input to the content-based Recommendation
algorithm.

58
A Comparative Study of Recommendation System used by Flipkart vs
Big Basket company:

In this competitive era of marketing, net advertising and marketing serves

as an powerful exchange change community that businesses can focus
directly to get direct entry to their customer goals.
Online buying facilitates customers to purchase goods and offerings proper
away from the merchants over the internet and without human
interference.
Consumers find products of their preference with just a click on of a
button. A whole collection of products from diverse dealers with their
costs get exhibited to the consumers for their choice. But, is competitive
fees and precise product functions enough for the consumers for their
shopping for selections.

This Comparative study between Flipkart vs Big Basket company

recommendation technique attempts to reply to this ambiguous situation
and try and understand the critical significance and advantages of their
appropriate recommendation filtering technique in their target audience.

OBJECTIVES:

● To understand the implementation of Recommendation system at

different sectors

● To understand tactics and methods that are used by e-commerce

players to grab the customers in India.

59
 ● To know how consumers are evaluating e-commerce sites for their
purchases on the basis

● To find out new opportunities after implementing different

recommendation technique/algorithm

Recommendation system employed by Flipkart and Big Basket

Filtering technique at Flipkart Filtering technique at BigBasket

Hybrid of content and collaborative filtering techniques Content-based Recommendation filtering technique

Recommendation type: “Similar product Recommendation Type: “Content-based

recommendation” Recommendation”

Procedure: Procedure:

The content matching is done over product attributes and In this Content-based approach, the user needs to
images in the catalog; and the collaborative filtering input information on their personal tastes, though
algorithm is applied over users' browse data (like on the same scale. The good side of the content-
product page views, Wishlist, add to cart, etc.) to find based approach is that it doesn't suffer from the
the most frequently co-browsed products for a given cold-start problem that even the first user can get
product. The ranked list of similar products based on the useful recommendations if the products are
relevancy is obtained from combining these multiple categorized and cataloged perfectly.
sources.

Applications: Applications:
When a customer of Flipkart researches through When a customer enters into their Bigbasket online
the catalogs of a product range, the search account, they need to check a huge list of items
patterns are recorded and a persona is created for even for a small set of required products for them

60
 the customer so that when he returns to the to select into their cart. The customers’ set of
website, the searching time is drastically reduced regular products to purchase is usually quite small
by showing the most relevant product that the when compared to the complete list of products in
customer could be interested in buying. the website. It takes a big effort and time of about
20 to 30 minutes to place an order as they are being
Another example, When a flip kart customer placed on mobile devices.
buys a mobile phone, out of the millions of
products in the catalog, the system correctly Results:
matches the most probable product you will In the above both situations, The content-based
purchase next - a mobile case or a screen approach needs high knowledge of huge inventory
protector – an example of content-based filtering. of Bigbasket items and Bigbasket has that huge
Also, the system can recommend products based data suitable for building a recommendation
on the person's persona or the persona of a algorithm. Each item must be characterized and
similar customer from the same demography. catalogued based on its characteristics
Results:
In the above both situations, There are two issues
faced by a customer. So, The appropriate So, The appropriate recommendation would be
recommendation would be Hybrid of content "Content-based Recommendation”
and collaborative filtering technique

Recommendation/Suggestions to Flipkart: Recommendation/suggestions to Bigbasket:

1. While combing two or more filtering technique in the

different approach in order to increase the accuracy level While building up a recommendation
and performance of the current RS implied, These algorithm/technique, particularly for content based

techniques harn their strength while leveling out their technique, it is critical to recollect NOT to upgrade

corresponding weakness just for a single measurement/user.In other words,

The following parameters are need to be keep in mind For Example

while developing a rs: The news news story suggestion to any news media

weighted hybrid, mixed hybrid, switching hybrid, application, it is not suggested to get a more

feature-combination hybrid, cascade hybrid, feature- notifications, Since this implies that we are

augmented hybrid and meta-level hybrid suggesting the clients the news/content they would

61
The problem of sparsity of data and cold-start was have normally viewed without our recommendation
addressed by combining the ratings, features and
demographic information about items in a cascade
hybrid recommendation technique
In case the Big basket diversifies to other segments
2.The company should upscale their filtering and expands their market then the company should
technique(Hybrid) with the new technologies through be prepared with various filtering techniques to
designing their own analytical engine for individual sustain their customer base for the particular
customer databases in order to overcome various segment as well.
challenges and issues faced when customers browse
their website.

62
Porter’s Five Forces framework analysis for E-commerce India:

Threat of Substitutes: Threat of substitutes is high as there are a lot of sellers with
similar products and services; and there is no switching cost for customers.
Bargaining Power of Suppliers:

63
Low – Bargaining power of suppliers is low as there are many suppliers in the market, and
therefore the ecommerce companies have the power to choose their suppliers.

Bargaining Power of Buyers:

High – Bargaining power of customers is very high as there are many players in the market
with similar products and there is no switching cost. Buyers prefer the company that offers
the best price among other factors.

Competitive Rivalry:
High – Competition among major players is very high, as there is no switching cost for
customers. The players are constantly competing on the basis on price as well as other factors
that influence buyers’ choice like quick delivery, discounts and offers, variety, customer
service etc

Threat of New Entrants:

High – Threat of new entrants is high, as there is very little cost involved in setting up an
ecommerce website.

Strategies Adopted by Indian E-commerce:

Expansion :
E-commerce companies are gradually expanding to different cities, regions and even
countries. They are also expanding their product range to cater to a larger number of people.

In May 2017, Uber launched UberEats, an on-demand food delivery app in India

As of July 29, 2017 Amazon India has pledged US$ 500 million for the expansion of its food
retail business in India.

Paytm Mall, ecommerce platform of Paytm, is planning to expand its groceries segment and
is targeting a Gross Merchandise Value (GMV) of US$ 3 billion from this segment by the
end of 2018.

Ancillary services:

64
One of the biggest advantages of E-commerce is that along with the core product or service it
can also provide numerous ancillary services without having to invest a lot.

Amazon India introduced a one day delivery guarantee on purchases for a nominal fee.
Flipkart started with same day guarantee in 2013 too.

Flipkart introduced its own payment gateway Payzippy and also its own logistics and supply
chain firm Ekart. E-commerce websites are also introducing e-Wallet services; for example -
Amazon’s Pay Balance.

Personalised Experience:

Site visitors demand one-of-a-kind experiences that cater to their needs and interests.

Technology is available, even to smaller players, to capture individual shoppers’ interests and
preferences and generate a product selection and shopping experience led by individualised
promotions tailored to them.

Many E-commerce websites provide personalised experience to customers to cater to their

needs and interests depending upon their location, choices, products they like or buy,
websites they visit etc. This strategy has helped companies to know customers’ demands
better and serve them accordingly

Subscription for ecommerce:

● E-commerce companies are increasingly adopting a subscription model to provide

extra benefits and tailored services to customers to suit their needs.

● Amazon introduced Amazon Prime, a subscription based service for Amazon

customers, in 2016. members of Amazon Prime could avail early access to selected
deals, free one day delivery and other benefits.

65
● Amazon Prime subscribers in India stood at around 5-6 million as of December 2016
In 2014, Flipkart introduced Flipkart First, a premium subscription based services
wherein a customer gets free delivery, discounted same day delivery, priority
customer service etc

66
Flipkart SWOT analysis

Strengths in the SWOT analysis of Flipkart

● India’s Largest E-commerce Retailer: Flipkart is the India’s largest E-commerce

company & had sold GMV (gross merchandising value) of $1 billion till now.
● Experienced founders: The Founders of Flipkart, Sachin & Binny bansal are Ex-
Amazon employees. Having prior experience in the E-commerce industry helped the
founders to work strategically and differentiate their business in a highly competitive
market.
● Acquisition: With its series of acquisitions like Letsbuy.co,, chakpak.com,
weread.com, Mine360 & the recent one Myntra in 2014 has helped the company to
expand in the E-commerce space & used the capabilities and existing resources of
acquired companies.
● High Brand recall: Flipkart has established itself as a renowned E-commerce
company in India through TV ads, online branding and through its presence on social
media. Brand activities like the “Big billion day” have really increased the brand
recall of the company.
● Own Payment gateway & Logistic arm: Having its own Logistics arm Ekart &
payment gateway Payzippy has helped the company to control its Expenses. Thereby
passing the benefits to the end customers.
● Exclusive & broad range of products: From having Exclusive rights to launch some
products like MotoG MotoX, Xiaomi Mi3 as well as personal designers segments in
garments category, has helped the company to differentiate and localize its offerings

Weaknesses in the SWOT analysis of Flipkart

67
 ● Limited Distribution channel reach: Although its logistics arm has kept
cost’s low, the reach has been affected which is a weakness for Flipkart.
Due to the use of outsourcing, Global giants like Amazon & eBay can
deliver the product anywhere in the country. However, Flipkart is still
struggling in this field.
● Cost of Acquisition: Due to stiff competition in the market & low customer
retention, the cost of Acquisition is high because Flipkart acquires a lot of
customers through online advertising. As per Flipkart data, the company
spends R.s 400/- on acquiring a new customer on an average.
● Power in the hand of buyers: Since this industry is flooded with many
players, buyers have a lot of options to choose. Switching costs are also less
for customers since they can easily switch a service from one online retail
company to another. Same products will be displayed in several online retail
websites. Product differentiation is almost absent and the fight then begins
on the basis of price only.

Opportunities in the SWOT analysis of Flipkart

● Expansion of business: By targeting other emerging markets

companies can increase their revenues as well as it can have
Economies of scale.
● Expanding their Product categories: This will increase their customer
base & at the same time will reduce the cost of acquisition and
customer switch.
● Changing mentality of Indian customers: With increasing numbers of
customers getting comfortable with online shopping & increase in
numbers of Internet users in India, there is huge potential in this
Industry.

68
 ● Supply chain: By optimizing their supply chain they can compete with
the other players & can manage the losing sales on account of not
making the product available due to delivery constraints.
● Established in other developing economies: Like Amazon, Flipkart can
slowly start expanding out of India and establish operations in other
countries as well which will help improve revenues.

Threats in the SWOT analysis of Flipkart

● Competition: Stiff competition from the global players like Amazon,

eBay as well as local players like Snapdeal, Tolexo and Shopclues
who are continuously trying to eat each other’s market share.
● Government regulations on the issues related to FDI in multi branding
retail has been a big hurdle in the success of the E-commerce industry
in India.

SWOT Analysis of Big Basket

Strengths in the SWOT Analysis of Big Basket :

● Product Range: Big Basket offers a huge range of products which

stretches to about 18000 products.
● Exotic Range: It also offers exotic fruits, vegetables and imported
groceries which are not easily available in nearby retail outlets. This
gives them an edge in catering to customers who like buying such
stuff.

69
 ● Discounts: Big Basket offers huge discounts on various products. They
also have the product bundling technique to maximise sales and offer
maximum discounts to the customers. This makes them a lucrative
option for customers to buy.
● Convenience: Big Basket offers an escape from the hassle of standing
in the queues of malls or going all the way to the retail stores. They
give the products right at the doorstep and offer multiple payment
options. The on-time delivery guarantee makes them a lucrative
option. Customers, especially in big cities where life is very fast and it
is difficult to reach thThe customers, can shop at any time of the day.
● Target Group: They have selected their target customers very carefully
and have established a footing in the metropolitan and some tier-I
cities. They operate in Bangalore, Hyderabad, Mumbai, Pune,
Chennai, Delhi, Mysore, Vadodara, Patna, Indore, Vijayawada, etc.
They are currently operational in about 25 cities.
● Low Fixed Cost business model: Most of the items that Big Basket
offers are perishable hence they depend on the retail stores they are
coordinating with for the products. This means they have no inventory
cost and makes their business model more profitable.

Weakness in the SWOT Analysis of Big Basket :

● Delivery Time: Big Basket delivers the groceries the next day. Many
retailers have started the facility of home delivery to nearby societies and
customers in just a few minutes to hours. So customers prefer this instead of
waiting for 24 hours.
● High Variable Cost: Cost of running many delivery guys, delivery trucks,
storage for perishable items, wastage during transportation makes the business
run at a high variable cost. They have to bleed more money and will take longer
to break even.

70
 ● Minimum order quantity/price compulsion: Big Basket does not offer home-
delivery below a certain set order price limit. This means that customers would
be forced to add a product or two just to avail the service. This makes them lose
customers.
● Be at home: Customers have to be at home when Big Basket is about to deliver.
This means the customers have to plan according to the delivery time. If by
some means the order gets delayed, the customer gets angry and unhappy by the
service. On top of this they have to pay a little add on price for the home
delivery.
● Order Cancellations: They depend on stores that they have tied-up with to
provide them the items. If they do not have the items available they won’t be
able to deliver them. This sometimes tends to result in cancellation of orders.
Out of stock issues tend to make the customers unhappy and repetitive of such
issues makes the customers reluctant to reuse the service.

Opportunities in the SWOT Analysis of Big Basket :

● Market: Indian retail market is valued at more than approx. $550 million and
grocery alone covers 60% of this share. Hence it’s a huge market with many
players to accommodate and customers to cater.
● Grocery Growth: India is the 6th largest grocery market in the world. E-
commerce grocery market is growing with a year on year growth rate of 19%.
● Expansion: Many cities are still not covered by these startups and Big Basket
can target them and gain the first movers advantage. They can expand to tier-2
cities and increase their market.
● Cross selling and Value Packaging – Cross selling is a major advantage for Big
basket as it can sell multiple products by product bundling.

Threats in the SWOT Analysis of Big Basket :

71
● Competition: It has a small presence in terms of cities covered hence,
it is getting a huge competition from other startups like Grofers,
PepperTap, Nature’s Basket, ZopNow, Aaram Shop, Mera Grocer,
etc.
● Bigger Players: Bigger players in terms of financial backing and
presence are also entering this lucrative industry. Players like
Amazon, Flipkart and Google have also entered this industry. This
may prove fatal for smaller players like Big Basket to compete with
them.
● Smaller Players: Many other localized shops have picked up on this
trend and have started home delivery service to nearby customers
thereby killing groups of target customers across regions and cities.
● Customer Retention: It is very difficult to retain customers. They
would tend to move with the service provider that offers the most
discounts.

CHAPTER 4
72
FINDINGS AND CONCLUSION

The study's key results, grouped based on criteria set out in the study's objectives, were listed
below:

● Flipkart being a market leader uses one of the best recommendation models, i.e Hybrid
and collaborative filtering which retain their customer at different purchasing behaviour
for a longer period of time.

● Since Big Basket has lesser market share in indian e-commerce industry i, which is
limited to the groceries segment but still able to handle their customer base through
content based filtering techniques.

● The comparative study of Recommendation system at Flipkart and Big Basket has able
to differentiate to different customer situation and the post-implementation outcomes of
the technique

● The role of Recommender system at e-commerce companies for extracting value for a
business from its databases, These technology help users as an online shopping
assistant to find items whatever they want to buy from a company.
● Recommendations system helping e-commerce companies by generating more sales
and also giving additional revenue and becoming a crucial tool to the company
● With the help of Recommendation technology, e-commerce and other online businesses
are being strengthened by huge volumes of customer data which help analysing their
purchase taste to take the managerial decisions of the company.

● The implementation of new technologies at RS can dramatically improve the scalability

of the recommender system.

● Recommender system enables users to find things that are interesting and it narrows
down the set of choices .It also provides additional and unique personalised service for
the customer.For Example, Customization per client/user.

● In the longer run, It increases customer trust and loyalty towards the company.

73
● It increases the customer engagement to the website and helps and assists the customer
as the comparison engine through comparing the products with different companies and
The companies also obtain more knowledge about the customers.

● The entire research helps in experimentally evaluating a new algorithm to issue faced
by users while online shopping through CF based recommender systems and results
above shown that item based techniques allow CF-based algorithms to scale to large
data sets and at the same time produce high-quality recommendations.

74
REFERENCES

E-Books

https://sdmimd.moodle.school/pluginfile.php/4706/mod_resource/content/1/1.%20About%20the
%20course.pdf
https://sdmimd.moodle.school/pluginfile.php/4707/mod_resource/content/1/2.%20Managerial
%20Decision%20Making%20and%20the%20role%20of%20Analytics.pdf

Reports

S. Shaikh, S. Rathi and P. Janrao, "Recommendation System in E-Commerce

Websites: A Graph Based Approached," 2017 IEEE 7th International Advance
Computing Conference (IACC), Hyderabad, 2017, pp. 931-934, doi:
10.1109/IACC.2017.0189.

Articles and Websites

https://en.wikipedia.org/wiki/Flipkartv

www.mygreatlearning.com

www.analyticssteps.com

labs.eleks.com
www.expresscomputer.in
www.emerald.com
journalstd.com
www.bigbasket.com
www.snapdeal.com
www.myntra.com
www.livemint.com
https://www.livemint.com/Industry/ZJ3AbUNvtPEb5nJ9ZER8WN/Using-Big-Data-to-net-online-
buyers.html
towardsdatascience.com
www.sciencedirect.com

75
www.mygreatlearning.com
ieeexplore.ieee.org

76