PLEASE DO NOT LEAVE ME!
Presentation Master thesis
Koen Schuurman
RESEARCH QUESTION
What are the effects of including customer heterogeneity for
estimating customer churn probabilities in a non-contractual
online retail setting?
0,187︎ 0,230︎ 0,150︎ 0,160︎ 0,170︎ 0,180︎ 0,190︎ 0,200︎ 0,210︎ 0,220︎ 0,230︎ 0,240︎
Standard model︎ Model with customer heterogeneity︎
Model fit R2
Growth:
18,3%
WHY THIS TOPIC
• ‘Acquiring a new customer is 5 to 25 times more expensive than retaining
an existing one’ (Kotler, 2001)
• Customer churn
• ‘The probability that a customer leaves the firm in a given period’
• non-contractual setting; suffers from the problem that customers have the
opportunity to continuously change their purchase behaviour without informing the company about it (Buckinx & van den Poel, 2005)
• The rise of the Internet
• Easy to obtain information
• Offers a wide range of choice options
• Differences in needs and preferences; expected to meet the needs and preferences of the customer.
• Customer heterogeneity
• Viewing a heterogeneous market as a number of smaller homogenous markets (Smith,1956; Wedel & Kamakura, 2001)
METHOD
•
Data
• 5000 customers randomly selected
• Period: 1st May 2015 – 1st May 2017
• Interpurchase time = churn
•
Ordinary least squares regression (standard model)
•
Latent class analysis
• Customers within the same latent class are homogeneous on certain
criteria; while customers in different latent classes are dissimilar from each other in certain important ways.
•
Data cleaning
RESULTS
•
Standard model; model fit (R2) = 0,187
• 8 hypothesis are supported by the results
• 4 hypothesis showed a reverse effect
• 4 hypothesis are rejected
•
Latent class regression model; model fit (R2) = 0,230
•
4 segments
1. ‘Together we dress to impress’
2. ‘Vintage will always be hip and happening’
3. ‘The Fashionistas’
DISCUSSION
•
Analysis of churners in the dataset
• Always an arbitrary choice
•
Reverse effects
• Return ratio; expectancy disconfirmation theory
• Age: lack of experience prevents them evaluating online shopping
•
Including customer heterogeneity improves model fit
• Do not treat all customers in an equal manner, because the variation
CONTRIBUTION
•
Theoretical
• Including customer heterogeneity improves model fit
• Drivers of non-contractual churn
• Interpurchase time as appropriate measure for customer churn
•
Managerial
• Increase the chances to indentify potential customers who might leave the
organization
• Better and more data driven policy regarding building sustainable
LIMITATION & FURTHER RESEARCH
•
Limitation
• Fashion data only
• Not a balanced set
• Partial churn
• ‘Correlation does not imply causation’
•
Further research
• GMOK model including dynamics and customer heterogeneity
• Run the same procedure for other product categories
