0% found this document useful (0 votes)
27 views18 pages

Week 10

The document discusses RFM (Recency, Frequency, Monetary value) analysis as a customer segmentation strategy, highlighting its significance in identifying high-value customers, personalizing marketing campaigns, and predicting churn. It explains the importance of each RFM component, methods for scoring and segmentation, and the use of K-means clustering for analyzing customer behavior. Additionally, it provides a case study of a retail company, TrendHub, illustrating the practical application of RFM analysis in optimizing marketing strategies and improving customer engagement.

Uploaded by

Riya singh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
27 views18 pages

Week 10

The document discusses RFM (Recency, Frequency, Monetary value) analysis as a customer segmentation strategy, highlighting its significance in identifying high-value customers, personalizing marketing campaigns, and predicting churn. It explains the importance of each RFM component, methods for scoring and segmentation, and the use of K-means clustering for analyzing customer behavior. Additionally, it provides a case study of a retail company, TrendHub, illustrating the practical application of RFM analysis in optimizing marketing strategies and improving customer engagement.

Uploaded by

Riya singh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

BUSINESS INTELLIGENCE & ANALYTICS

RFM ANALYSIS
COURSE INSTRUCTOR: PROF SAJI K MATHEW

K R Subisha
TA – BI&A
PhD Scholar, Information Systems
IIT MADRAS
INTRODUCTION
• RFM stands for Recency, Frequency, and Monetary value
• Traditionally used for customer segmentation strategy
ØRECENCY: Amount of elapsed time since last purchase
ØFREQUENCY: Count of purchases done during a time period
ØMONETARY VALUE: Total amount of purchases made during a time period

• Customers who rank high in RFM are more likely to be of high value
• RFM doesn’t require complex statistical analysis or detailed demographic data
ØHelpful in predictive modelling techniques like churn modelling
ØUseful for targeting mails
SIGNIFICANCE OF RFM
• Identifying High-Value Customers: Customers with high scores in all three categories
(Recency, Frequency, Monetary) are likely to be the most loyal and profitable.

• Personalized Marketing Campaigns: By segmenting customers based on their RFM


scores, businesses can tailor marketing campaigns to different customer groups.

• Retention and Churn Prediction: RFM analysis helps identify customers who are at risk of
churning. With declining recency, frequency, or monetary value, companies can
proactively address their needs and reduce churn.

• Product Recommendations: RFM is useful in profiling customers to make relevant


product recommendations and upsell or cross-sell opportunities.

• Optimizing Resources: RFM segmentation allows businesses to allocate resources more


efficiently, it doesn’t require demographic or psychographic data.
RFM and CLV
• RFM along with Customer Lifetime Value (CLV) analysis could provide actionable insights to
Marketing decisions
• CLV is the predicted net profit a company can expect to earn from a customer over a future
horizon
ØHigh R, F, and M: Most valuable and engaged customers, may reward them with loyalty
programs or exclusive offers to maintain their loyalty and increase their CLV
ØHigh R but low F, and M: Need to be re-engaged to make more frequent and higher-value
purchases to boost their CLV
ØHigh F but low R, and M: Requires incentives to increase their spending and engagement,
to ultimately raise their CLV
ØLow R, F, and M: Customers might not be profitable. Companies should decide if it's
worth investing in re-engaging them or if they should focus on more promising customer
segments.
“ We’re going to look at the flow of transactions, the size of transactions, other interactions that we’ve had with each and every customer, and we’re
going to make an educated guess about what we think that customer will be worth in the future ”
-Wharton marketing professor Peter Fader about Delta Airlines
WHICH IS MORE IMPORTANT?
• Recency is considered the most important of the three variables R, F and M in
predicting future behavior .
ØMany direct marketers base their contact decisions mainly on recency

• Customers often fall into two groups: "hot" and "dead," and the R score will reflect
this

• Recent customer behavior is a strong indicator of their current level of engagement


with a business.

• Customers who have interacted or made purchases recently are more likely to be
actively interested and engaged with the products or services of a business.

Miglautsch, J. R. (2000). Thoughts on RFM scoring. Journal of Database Marketing & Customer Strategy Management, 8, 67-72.
WEIGHTING AND COMPOSITE SCORING
• RFM scoring is typically based on life-to-date totals, considering a customer's entire history.

• Weighting RFM scores is useful for comprehensive customer segmentation .

• Various methods for combining R, F, and M values are present, allowing for more refined
customer scoring and segmentation

• Some mailers multiply R by 5, F by 2, and M by 1 to create composite scores.


ØThis approach emphasizes recency and frequency, giving more importance to the most
recent customers.

• Different scoring methods have their strengths and weaknesses, and it's important to choose
the most suitable method for effective customer segmentation and marketing.

Miglautsch, J. R. (2000). Thoughts on RFM scoring. Journal of Database Marketing & Customer Strategy Management, 8, 67-72.
ASSUMPTIONS
• Many businesses assume that loyal customers are less costly to serve, willing to pay more,
and effective at promoting the company through word-of-mouth recommendations.

• There can be four distinct groups, not just two (loyal and disloyal) which are Butterflies, True
friends, Barnacles, and Strangers.

1.Butterflies: Profitable but disloyal customers. They may make purchases, but they are not
committed to the brand. The challenge with butterflies is to maximize revenue from them while
they are buying.
2.True Friends: Highly loyal and profitable customers. They are likely to remain loyal to the brand.
Businesses should take a softer, relationship-focused approach with true friends to maintain and
enhance their loyalty.
3.Barnacles: Highly loyal but not very profitable customers. They may have untapped potential for
increased spending. Identifying how to encourage barnacles to spend more is a priority.
4.Strangers: Customers who generate neither loyalty nor profits. The recommendation is to
identify strangers early and avoid investing resources in them.

Reinartz, W., & Kumar (2002). The mismanagement of customer loyalty. Harvard business review, 80(7), 86-94.
RFM FOR CUSTOMER DONATIONS
• RFM offers useful insights to analyze a donor base
Ø Take common problems of LYBUNTs or SYBUNTs — donors who gave “last year but
unfortunately not this” year or “some years ago but unfortunately not this.”
• RFM scores in a 1-5 scale helps an organization target lapsed-donors on the 1-4-5s
(high-value donors who used to give frequently) rather than the 1-2-1s.
• Monitoring how donors make payments, we can consider promoting credit card
options to enhance donor engagement and turn some 4-2-3s into 4-5-3s.
• With experience, RFM analysis can be expanded and combined with detailed donor
demographics to gain new insights and improve results in direct mail fundraising
campaigns.

Blackbaud. (June 2004). RFM: A Formula for Greater Direct Mail Success. Blackbaud Inc
TRENDHUB: A Case Of Customer
Segmentation using RFM
A retail company called "TrendHub" wants to conduct an RFM analysis based on their raw data.
The dataset includes information about various aspects of their sales transactions, such as
product details, customer information, pricing, and logistics. TrendHub operates several stores,
each identified by its IssuingPlantCode. They sell a variety of branded products (Brand) across
different styles, colors, and sizes. The transactions include sales (TransType) with transaction
prices (TranPrice) and quantities (TrnQty). Each sale is associated with a timestamp (TrnTime).
Additionally, tax details, discounts, and transporter information are included.

The company is interested in RFM analysis to categorize its customers based on Recency,
Frequency, and Monetary value. They want to understand which customers are the most active
and valuable, enabling them to tailor marketing strategies effectively, reward loyal customers
(Loyalty Points), and improve their overall sales performance. This analysis will provide valuable
insights for TrendHub's future business decisions and customer engagement strategies.
DATA CLEANING AND DATA PREPARATION
• Treating the outliers
Ø Transaction price can never be negativeà removed all the records with a negative Transaction price.
Ø Transaction Quantity can never be negativeà removed all the rows with a negative Transaction Quantity.

• Returns are considered as negative values since it has to be subtracted if the customer returns
the product

• Pivot tables are applied in excel and the following are obtained:
Ø Sum of total value gives the total monetary transactions performed by a particular custom (both sales and
returns)-MONETARY
Ø Count of transactions quantity gives the total number of visits the of customer to the store- FREQUENCY

Ø Days since last order(Current date-most recent transaction) -RECENCY


GROUPING AND SCALING
• Since R,F,M values have to be scaled for better efficiency during clustering and ease of
comparative analysis, we scale it into values from 1-10
• Recency value obtained by grouping Days since last order into 10 deciles and then scaling
them from 1-10
• Reverse scaling done for Recency since the most recent customer needs to be given the
highest R value
• Frequency values have been obtained by grouping Count of txn qty into deciles and then
scaling it to 1-10
• Monetary values have been obtained by grouping and scaling Sum of total value column
• RFM=R*F*M
• Next step is customer segmentation which will be done with this preprocessed data in
python
NTILE FUNCTION IN SQL
• The SQL NTILE function is a valuable tool for RFM (Recency, Frequency, Monetary) analysis,
aiding in segmenting and ranking customers into distinct groups based on their behavior.
• This function divides a dataset into equal parts, creating quantiles or percentiles, which can
be very useful for categorizing customers according to their RFM scores.

CREATE TABLE RFM AS SELECT customerid, recencydays, NTILE(5) OVER (ORDER BY


recencydays) AS Recency FROM Transactions;

• The NTILE function assigns a numerical ranking to customers within their respective
segments, reflecting their RFM scores in comparison to others.
ØA customer in the "top 20%" segment will have a higher NTILE rank than one in the
"bottom 20%."
RFM ANALYSIS
• Sort recency, frequency and monetary values in one of the 10 buckets according to priority
• Reverse scaling needed for recency value
BUSINESS INTELLIGENCE & ANALYTICS

K MEANS CLUSTERING OF RFM


COURSE INSTRUCTOR: PROF SAJI K MATHEW

K R Subisha
TA – BI&A
PhD Scholar, Information Systems
IIT MADRAS
K-MEANS CLUSTERING

Once customers are assigned RFM behavior scores, they can be grouped into segments
and their subsequent profitability analysed. This profitability analysis then forms the
basis for future customer contact frequency decisions.
1. Customer Segmentation: K-means clustering is a data-driven method for grouping
customers based on behavior or characteristics, revealing valuable segments not easily
identified through manual categorization.
2. Unbiased Analysis: K-means clustering is criterion-independent, analyzing the entire
dataset to unveil hidden patterns in large and complex data.
3. Productand Content Recommendations: K-means clustering enables businesses to
understand customer segments, leading to personalized recommendations and content
that enhance the customer experience and boost sales.
ANALYSIS
Cluster 0: STRANGERS Cluster 2: BUTTERFLIES
ØRecency (Mean, Std): 1.57, 0.99 ØRecency (Mean, Std): 1.77, 1.09
Ø Frequency (Mean, Std): 3.88, 2.09 ØFrequency (Mean, Std): 8.24, 1.6
ØMonetary (Mean, Std): 4.46, 1.89 ØMonetary (Mean, Std): 8.29, 1.53

Cluster 1: TRUE FRIENDS Cluster 3: BARNACLES


ØRecency (Mean, Std): 7.18, 1.77 ØRecency (Mean, Std): 6.35, 1.95

ØFrequency (Mean, Std): 8.38, 1.6 ØFrequency (Mean, Std): 2.40, 1.76

Ø Monetary (Mean, Std): 8.6, 1.56 ØMonetary (Mean, Std): 3.34, 1.84
KEY OBSERVATIONS
1. Strangers(Cluster 0): These customers exhibit low recency, moderate frequency, and
moderate monetary values. They are recent customers who make purchases occasionally
but not at high value.
2. True
Friends (Cluster 1): True Friends have high recency, high frequency, and high monetary
values, suggesting they are loyal customers who frequently make significant purchases.
They represent the most valuable group.
3. Butterflies
(Cluster 2): Butterflies have low recency, high frequency, and high monetary
values. They are profitable but not highly loyal, making occasional high-value purchases.
4. Barnacles
(Cluster 3): Barnacles have moderate recency, low frequency, and low monetary
values. They are moderately loyal but not very profitable, indicating potential for increased
spending.
These clusters provide valuable insights for marketing strategies. True Friends are the most
valuable, and strategies should aim to retain them. Butterflies may require tactics to boost
loyalty, while Barnacles may benefit from efforts to increase their spending. Strangers may
not be a significant focus due to their low contribution.
THANK YOU

You might also like