Customer Segmentation Using RFM Analysis
Customer Segmentation Using RFM Analysis
Customer Segmentation Using RFM Analysis
24 www.pbr.co.in
Pacific Business Review International
profitable offerings as a part of loyalty programs. Also, it is possible to focus on planning for customer retention programs.
Originality/Value: The value of this research paper lies in realizing through mapping the theoretical value of RFM analysis
to practically visualizing the customer segmentation in the form of grouping best customers and low-risk customers using
Python implementation. Such value can be found in following the research methodology steps while developing the
content of the paper as well as showing practically obtaining results through dealing with data-based case study as a part of
serving itself as a testing the proposed hypotheses.
Keywords: Behavioral Segmentation, Clustering, Customer Segmentation, Electronic Customer Relationship
Management, Geographical Customer Segmentation, Market Segmentation, RFM Analysis.
www.pbr.co.in 25
Volume 13 issue 11 May 2021
behavior, who might require separate products or who may purchase patterns. These purchasing patterns are
respond differently to various combinations of marketing interpreted by purchasing portfolios, which include list of
efforts” (Ahmad, 2003) (Aksoy, Keiningham & Bejou, categories of customer purchases and their consumption
2014).It includes (i) geographical (ii) demographical (iii) behaviors on these categories (Raje & Srivastava, 2014).
behavioral and (iv) psycho graphical based segmentations. The rationality for categorization is the consideration of
Customer segmentation improves profits through direct factors such as percentage of frequency and percentage of
marketing where in catalogs, mailers etc are used (David monetary of all categories in range. It uses the clustering
Shepard Associates, 1998). analysis by K-Mean technique (Do, 2011). Susilo (2016)
conducted research on behavioral segmentation and price
Geographical Customer Segmentation
on customer value influenced for increasing the purchase
Location is a transcendental decision to ensure the viability that impacts the customer loyalty. The data is related to
of retail stores. Geo-demographic segmentation can be Jakarta. Huseynov & Ozkan (2017) have conducted the
considered as an analytical tool for optimizing retail research study on behavioral segmentation analysis of
location strategy. Geo-demographic segmentation seeks to online consumer audience in Turkey by using real-time e-
capture the spatial heterogeneity of the urban market areas commerce transaction data (Huseynov & Yildirim, 2017).
in terms of the characteristics of the residents (Gonzlez- Previous research in customer segmentation based on
Benito & Gonzlez-Benito, 2005). Geographical customer behavioral data in e-market place used Python
segmentation is basically segmenting customers based programming language to conduct cluster analysis and
geographical areas to which customers are belonging to visualization (Aziz, 2017). Further research on customer
(Sulekha & Mor, 2014). It is believed that consumers who segmentation based on user behavior analysis with RFM
live in the same region share similar wants and needs and model and data mining technique is conducted by Tavakoli,
are different from the consumers who live in other regions Molavi, Masoumi, Mobini, Etemad & Rahmani (2018),
of the world (Martin, 2011). In other words, the customers these authors used R programming for performing the
are divided based on geographical area units such as exploratory analysis and the RFM analysis is done using
villages, towns, cities, states, countries and regions. Store Python. Kamthania, Pahwa & Madhavan (2018), Yoseph,
wise segmentation divides a network of stores into Hashimah, Malim & Almalaily (2019); these people have
meaningful groups. Geographical categorization of visually shown the high valued and low valued customers.
customers is possible with data base techniques by Though there are few researches that used Python for RFM
employing a set of pre-categorized examples to develop analysis, our approach followed an approach that reduces
categorized population of records. the gap between theoretical RFM coding and practically
implementing such RFM coding based filtering data using
Previous research on geographical customer segmentation
clustering technique of data mining on an existing e-
using historic customer and customer sale transaction data
commerce transaction data of a retail management.
and filtering the data by customer location hierarchy
(Oracle docs, undated; Campbell, 2015) was seen much. In Customer benefit from behavioral segmentation by (i)
this research database filtering approach is used to filter the optimizing their market spending (ii) increasing the
customer transaction data in order to obtain the location customer lifetime value (CLV) (iii) improving customer
hierarchy such as country. Filtering customers based on service and customer experience (iv) implementing
customer country has been applied. optimal marketing channels selection for their each
segment (v) improving product features and offerings (vi)
Behavioral Customer Segmentation
identifying and cater to most profitable customers (Rivas,
Customer segmentation is a marketing strategy involving 2019).
the customer division into various groups based on their
Research Objective
underlying characteristics, needs and interests. It then opts
for designing and implementing marketing strategies to Target marketing is a method of focusing and attracting
target them. One of the most types of segmentation customers who are likely to buy a product. Data mining
approaches is behavioral segmentation analysis (Huseynov techniques can be used to identify the highly profitable,
& Ozkan, 2017). high-valued and low-risk customers (Pandey, 2017).
In behavioral segmentation, customer purchase behavior Clustering is a data mining technique. It is used for
needs are acknowledged not only in specific needs of geographical customer segmentation and consumer
products but also the interactions made among the whole behavior-based customer segmentation, in view of
range of products. This segmentation is done based on implementing target marketing strategy. In this research
26 www.pbr.co.in
Pacific Business Review International
the goal is: (i) how to implement customer segmentation of sort of occurring in a bounded context” as cited in Knoll
stores, based on multiple data sources and (ii) how to create (2008) in passing. Case studies are an established research
marketing strategies for each segment. approach for exploring complex phenomenon from such
domains where the extracting insights become hazy (Lovas
In this paper only (i) and (iii) consumer marketing
& Goshal, 2000).
techniques have been focused to be described and
discussed to be implementation using innovative tools and Due to the sensitive nature of primary data gathering,
languages like Python to realize the concept of consumer companies generally do not provide the primary data. As a
market segmentation. result, the required data is gathered from web sites in the
form of a secondary data, in order to process them in python
Research Hypotheses
programming and to realize the result of data mining
Null Hypothesis techniques (Schrantz, 2013). The secondary data is
collected during the period of May-June 2020 and the
H01 = Segmenting customers based on their geographical
research is conducted during May-July 2020.
residence is easy to work using Python programming kind
of innovative information technologies The methodology proposed in this paper contributes to the
electronic marketing literature by exploring analytical
H02 = Segmenting customers based on their purchase
Customer Relationship Management perspectives in the
behavior patterns of grocery items is easy to find using
retail management of grocery stores. This methodology
Python programming kind of innovative information
aims to make an attempt to identify customers and also
technologies.
segmenting them using clustering technique of data mining
Research Methodology through a case study as an exploratory method. It uses
secondary data available from web sites sources such as
Contemporarily technology based online retailing has
Kaggle.com and Github.com etc. Clustering analysis is
reshaped the retail landscape. Shopping in Grocery stores
mostly used as a data mining technique. It maps data items
is emerging as one of the fastest growing categories of
to more unknown similar item groups (Oliverira, 2012).
online retailing in the United Kingdom (Davies, Dolega &
Arribas-Bel, 2019). The development of the modern Realizing Geographical Segmentation in Python
patterns of grocery stores in Britain has been described and
Geographic segmentation variables can include location
analyzed as a part of case study (Guy, 1996).
hierarchy, city, state, country, population density,
This research uses a case study approach. It uses qualitative economic status, zip code and regional climate (Oracle
case study methodology to enable researchers in docs, undated). For example, if ecommerce ships
conducting deep exploration of complex phenomenon internationally, user interface may want drill down
within the specific domain context (Rashid, Rashid, behaviors, conversations and purchase patterns from a
Warraich, Sabir and Waseem, 2009). A case study approach particular geographic region or country or the location
is adopted for the investigation, when the realization hierarchy (Campbell, 2015).
through theoretical aspects becomes limited with
For Python Implementation of Geographical Segmentation
operational activities and synergies. According to Miles &
see Figure-1 to Figure-4.
Huberman (1994) a case study is a “phenomenon of some
www.pbr.co.in 27
Volume 13 issue 11 May 2021
28 www.pbr.co.in
Pacific Business Review International
www.pbr.co.in 29
Volume 13 issue 11 May 2021
30 www.pbr.co.in
Pacific Business Review International
www.pbr.co.in 31
Volume 13 issue 11 May 2021
Best way contributing Customers has to keep informing customers about new products, new
incentives, any loyalty programs or social media incentives
These are the customers with RFM score 5-5-5. It means
that is run by the stores.
these are the customers who bought recently, buy often and
spend a lot. It is likely that they will continue to do so. Since
they already like you, the recommendation is that the store
32 www.pbr.co.in
Pacific Business Review International
F=5 indicates that the customers are loyal with high potential at stores in terms of investing a lot in your
frequency of purchases at stores. But it does not mean they products. The recommendation here is that the stores that it
bought it recently and spent high money at stores in should keep informing about innovative products,
purchases. M=5 indicates that customers are big spenders. expensive products and top line products to this cluster
They spent a lot of money over the given transaction group of customers.
period; so, it is likely that their lifetime value is having
Least contributing Customers
These are the customers with RFM score 1-1-1. It means unlikely to be considered worth.
these customers last ordered a lot of time ago, bought very
Count of RFM Score
few times and customers spent very, very little. They are
www.pbr.co.in 33
Volume 13 issue 11 May 2021
Conclusion References
Customer market segmentation is an increasingly Ahmed, R. (2003). Benefit Segmentation: A Potentially
becoming significant issue in today's competitive useful technique of segmenting and targeting older
innovative marketing domain like retail groceries store consumers. International Journal of Market
management. In this paper, geographical segmentation is Research 45 (3), Retrieved on 04-11-2020
done by filtering the transaction data using data base https://www.warc.com/fulltext/JMRS/78268.htm
technique. Further, customer behavioral segmentation is
Ait Daoud, R. Bouikhalene, B. Amine, A. & Lbibb, R.
done based on RFM analysis using Python programming
(2015). Combining RFM model and clustering
statement constructs. Contemporarily, use of data mining
techniques for customer value analysis of a company
techniques is playing important roles in marketing. Though
selling online. 978-1-5090-0478-2/15, IEEE. [In
data mining techniques cannot fully replace the significant
text citation: Aid Daoud, Bouikhalene, Amine &
role of domain experts and their domain expertise
Lbibb, 2015]
knowledge, application of such techniques actively support
business experts in business oriented big data analytics and Aksoy, L., Keiningham, T. & Bejou, D. (2014). Profit
in making better decision in the area of target marketing Maximization through customer relationship
and electronic customer relationship management. Data management, Measurement, Prediction and
mining techniques can help organizations in extracting Implementation, New York: Routledge
patterns from the transaction data to gain a competitive
Aziz, A. (2017). Customer segmentation based on
advantage in their efforts to optimize customer relationship
behavioral data in e-marketplace, Thesis submitted
management. Use of clustering algorithm is one of the
to Department of Information Technology, Uppsala
ways to segment data set according to the gathered
University, Sweden.
similarities in view of developing insights on behavior-
based customer segmentation. Bagla, R. K. & Khan, J. (2017). Customers' expectations
and satisfaction with online food ordering portals.
Further, effective loyalty programs rely on sophisticated
Prabandhan: Indian Journal of Management, 10(11).
CRM analytics to understand factors that influence the
[Bagla & Khan, 2017]
consumer behavior. Understanding consumer behavior
allow to build loyalty programs by adding value in term of Banerjee, K. (2018). Enhancing customer loyalty using
conducting customer segmentation and product market basket analysis, big data and analytics
personalization (Zamil, Al Adwan & Vasista, 2020; customer experience, customer sentiment analysis,
Banerjee, 2018). enterprise analytics, market basket analysis, April 2,
Retrieved from Aspiresys.com blog. [In text citation:
Further the research study is limited in considering online
Banerjee, 2018]
customers, who order to have grocery items to be delivered
online. The research value can be observed on bridging the Berson, A., Smith, S. & Thearling, K. (1999). Building
gap between theoretical understanding and the visible Data Mining Applications for CRM, 1st Edition,
realization through implementation (Bagla & Khan, 2017). New York: McGraw-Hill Professional. [In text
As cited in Kasliwal & Agarwal (2014), based on the citation: Berson, Smith and Thearling, 1999]
modified theory of planned behavior, it is found that
Bilgic, E., Kantardzic, M. & Cakr, O. (2015). Retail Store
customers were ready to pay prices (Kim & Han, 2010)
segmentation for target marketing. In proceedings of
online when purchasing grocery items similar to those who
ICDM, Industrial conference on data mining:
make payment at brick and mortar stores without hesitating
Advances in data mining: Applications and
much.
theoretical aspects, 32-44. [In text citation: Bilgic,
Declaration of Conflicting Interests Kantardzic, Cakr, 2015]
The author(s) declared no potential conflicts of interest Bose, I. & Chen, X. (2009). Quantitative models for direct
with respect to the research, authorship, and/or publication marketing: A review from systems perspective,
of this article. European Journal of Operational Research, 195(1),
1-16. [Bose & Chen, 2009]
Funding
Campbell, K. (2015). Google Analytics: How to use
The author(s) received no financial support for the
Advanced Segments. Retrieved on 24-10-2020
research, authorship, and/or publication of this article.
https://www.practicalecommerce.com/Google-
34 www.pbr.co.in
Pacific Business Review International
www.pbr.co.in 35
Volume 13 issue 11 May 2021
36 www.pbr.co.in