Market Basket Analysis
Focusing on product-centric to customer-centric within the
lifecycle has emerged as a framework in the over-time stages; the relationship
between customers and retail business. Starting early in the customer lifecycle
is with limited data, but will gradually accumulate the amount of data from the
customer as the progress of the data is collected. Customer Development is
important because of these key factors:
·
Increase usage of current products
·
Sell additional (cross-sell) products
·
Sell higher margin (up-sell) products
·
Personalized communications and offers for
specific customers
·
Market-basket
analysis to identify opportunities for cross-promotion (Kenan-Flagler
Business School).
Market basket analysis is
commonly uses as three domains:
1.
The method of personalized recommendations
2.
Used as spatial distribution of items leading to more
sales
3.
The creation of marketing strategies focusing on
discounts and promotions based on customer’s behaviors via special sales or
target promotion (Ansari, 2019).
Application of this analytic method to
Customer Life Cycle stages:
1.
Prospect-
gaining new customers is about prospecting and searching for them. This can be
done by researching the completion’s customers and understanding their buying
habits. This also includes reaching marketing sales and promotions from the
competitor. The only way to gain new customers is either obtain them from
competition or from younger generation. Gaining new customers through market
basket analysis targets them by monitoring buying patterns and improving
customer services. Analyzing recurring patterns offering related items together
to increase sales. Once the company understands their behaviors they can send
out marketing promotions to lure them in. MBA (Market Basket Analysis) targets
customer baskets by monitoring buying patterns and improving customer services.
Analyzing recurring patterns offering related items together to increase sales
(Ansari, 2019).
2.
New Customer-
once the new customer is shopping at the new retailer, and has gathered the
previous data from the new customer. One way of tracking a customers’ buying
habits is having a rewards program giving them a specific ID to identify the
customer. Once the customer has applied to the program with a new account, the
retailer can use Market-basket analysis to determine their present and future
purchases of products sets. This means to study their buying habits of products
bought together! Also, a great way to analyze the product sets bought from the competition
and viewing if the same product sets are being bought in the new store or are
they buying “new product sets”? This small analysis will determine product
sales, personal recommendations and the intent of luring the new customer.
3.
Established
Customer- because these customers who have made more than initial purchases
over a time period, it is essential for them to remain as your customers based
by their loyalty. This is where you have great amount of data knowing the
products (first product bought and the second product they are likely to buy
with the first item) they buy frequently and what is their top items. Gaining
data for use in the MBA to assist in promotion and sales, and to be accurate as
possible knowing of their purchase. It would be beneficial to “promote
personalized items” to them in gaining more sales.
4.
Former Customer-
customers who leave; there can be an effort to regain their loyalty. Again,
there are ways to send promotions on personalized recommendations on their
favorite products or items they buy regularly. Depending on the reason of
leaving, attracting the customer back may not be easy, but based by the
promotion the company may offer on their “personal items” with great deals can
have a chance of bringing them back with enticement. A great way to do this
with MBA is promoting “Buy one, get one free” or save a certain amount on two of
their recommended items.
Explanation of Analytic
Method Used
1. How this method work
Market Basket Analysis
starts in retail organizations and targets customer shopping habits by
monitoring buying patterns and improves customer satisfaction. In monitoring
“customer’s buying habits’” relates to if the customer purchase item “A” would
question the likeliness of purchasing item “B”. This would determine
relationships between the products that were or are purchased together as
profiles of “if-then” rules (Svetina, & Zupancic, J., 2005 and Rstat).
2.
Strengths-identifies
as products purchased in pairs and of co-occurrence. Example if a customer buys
peanut butter (A), what is the likelihood the customer would buy bread (B), or
jelly (C) and the confidence level that comes with this data.
3.
Weaknesses- when using standard ARM (Association Rule
Mining) algorithms that are capable of identifying distinct patterns from a
dataset, fail to associate user objectives and business values with outcomes of
the ARM analysis. Another issues arises during data mining process to treat
data containing temporal information. Temporal Association Rule can add time
constraint on association rule (Pillai, 2011).
4.
When to use:
Use when seeking resources in learning product associations and bases of a
retailer’s promotion strategy on the product. Examples:
a.
Associating additional brands when sold together with
basic brands with increasing revenue.
b.
Campaigns of “Buy two, get three” sales in determining
the right product to promote.
c.
“Buy a product, get a gift” to determine the sales of a
product and gift relationship and high margin rate.
d.
Inquires relating to sets of products are well-defined
and sold together with a discount (Svetina, & Zupancic, 2005),
Alternative and
Complementary Methods
1.
Alternative
method- Time Series Clustering: clustering on sales transactions data
formatted as time series as an alternative to Market Basket Analysis because:
a.
Data matrix required by an association analysis becomes
large where there are many sales transactions and products and within each
transaction involves sales of only a few products.
b.
Time series clustering of sales connections requires
the data to be précised as time series resulting as a smaller data set
involving less time to process.
c.
Identifies products that are commonly purchased across
certain time periods.
When quantities of two
products are regularly purchased, the sales of two products are detected over
different periods in time. Time series of the sales quantities should be
comparable in their upwards and downwards temporal patterns. At the time that
clustering is applied, the two time series would be allocated to the same
cluster, which contains a set of comparable to an item set in association rule
mining (Tan, & Lau).
2.
Complementary
method- ARM (Association Rule Mining) rules in data mining research are the
findings of relationships between data items in large datasets. User-Centric
approach to MBA discovers the association or rules pertaining to frequently
occurring items with considering the benefits of the item sets to be equal. Within
real world datasets containing both frequent and infrequent or rarely occurring
items; will help in decision-making process (Pillai, 2011). ARM is most suitable method for analysis of
large market basket data containing large volume of sales transactions with
high number of products. These rules provides information such as “If-Then”
statements, which are commonly used to recognize the presence, nature and
strength of an association rule (Ansari, 2019).
Ansari, S., Z. (2019)
MARKET BASKET ANALYSIS: Trend Analysis of Association Rules in Different Time
Periods. NOVA Information Management School. Retrieved from https://run.unl.pt/bitstream/10362/80955/1/TEGI0458.pdf
Berry, M. and Linoff,
G. S. (2011). Data Mining Techniques: For Marketing, Sales, and Customer
Relationship Management, 3rd
Edition. John Wiley & Sons P&T, 2011. VitalBook file
Kenan-Flagler Business
School. The Customer Lifecycle. The University of North Carolina. Retrieved
from https://csbweb01.uncw.edu/people/howe/classes/mba541/Customer_lifecycle.pdf
Pillai, J., (2011) User
centric approach to itemset utility mining in Marketing Basket Analysis. Bhilai
Intsitue Of technology Retrieved from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.301.3985&rep=rep1&type=pdf
Rstat, Explanation of
the Market Basket Model, Retrieved from https://webfocusinfocenter.informationbuilders.com/wfappent/TLs/TL_rstat/source/marketbasket49.htm
Tan, S., C., & Lau
J., P., S., Time Series Clustering: A Superior Alternative for Market Basket
Analysis. Retrieved from
Comments
Post a Comment