1 | 28-01-2020
MAPPING THE CUSTOMER
JOURNEY AND IDENTIFYING
PROFIT DRIVERS
USING A LATENT CLASS CLUSTER- AND REGRESSION ANALYSIS
Table of contents
I.
Introduction
II. Theoretical framework
III. Research design
IV. Results
I. Introduction
› Nowadays: many users shop online, leading to an increase in the number of online customer journeys
› However: customer journeys are becoming increasingly complex
› In order to design effective strategies, one should know who their audience is and how they behave
› Mapping the customer journey has not been extensively addressed in research
RQ 1
Which customer segments can be identified by their use of specific touchpoints in the customer journey?
RQ 2
II. Theoretical framework
› Literature on touchpoints/ customer journeys/ behaviour as a segmentation base
› Segmentation literature on online customer behaviour
▪ Transactional data ▪ Longer period of time
▪ Multiple devices & channels
› Discounted Expected Transactions (DET) model
▪ Perfectly fits the available data ▪ Focuses on future transactions
III. Research design
RQ 1. Which customer segments can be identified by their use of specific touchpoints
in the customer journey?
› Segmentation using a Latent Class Cluster Analysis + Generates statistically consistent criterion
+ Different scale types of the variables can be used
+ Probabilities are computed from estimated parameters and observed scores › Multiple information criteria used for cluster selection
▪ Akaike Information Criterion (AIC) ▪ Bayesian Information Criterion (BIC)
III. Research design
RQ 2. What are the most important drivers for conversion for each of the newly created
segments?
› Regression analysis performed
› Dependent variable: Discounted Expected Transactions
1. Data transformation 2. Parameter estimation 3. Goodness-of-fit testing
IV. Results - segmentation
› Negative BIC
IV. Results - segmentation
› Homogeneity of variance might be affected
▪ Fligner-Killeen test performed,
all values below 0.05 › Kruskal-Wallis analysis of
variance
▪ Are the variables significantly
different?
› Paired Wilcoxon rank-sum test
▪ Do the segments per variable
IV. Results
Segment
TP interactions
CJ length
Transaction avg
1: Short users
Low
Short
0.163
2: Heavy users
Highest
Longest
1.263
3: Regular users
High
Long
0.558
IV. Results - DET
› Frequency: number of transactions made in the calibration period › Recency: date of most recent transactions
IV. Results - Regression
› No multicollinearity
› Count data
▪ Either Poisson- or Negative Binomial regression
› Pearson Chi-square dispersion statistic
IV. Results - Regression
› Antilog
V. Discussion - Managerial
› Better customer behaviour knowledge
▪ Able to better allocate budget towards significant touchpoints
› Need for variety
▪ Develop an application
› Customer journey length negatively influences number of purchases made
▪ Targeted advertising and better recommendations to drive users towards a
V. Discussion - Theoretical
› Contributed to knowledge gap of segmenting customers based on actual online touchpoint behaviour
› Proposed behavioural drivers towards profit
V. Discussion – Future research
› Increase external validity
▪ Add other countries ▪ Add other industries
› Methodological weakness of using behavioural data as a segmentation base
▪ Use surveys
VI. References
› Barnes, Stuart, J., Hans H. Bauer, Marcus M. Neumann, and Frank Huber (2007), "Segmenting cyberspace: a customer typology for the internet", European Journal of Marketing, 41 (1/2), pp. 71-93.
› Berk, Richard and John M. MacDonald (2008), “Overdispersion and Poisson Regression”, Journal of Quantitative Criminology, 24 (3), 269-84.
› Fligner, Michael A. and Timothy J. Killeen (1976), “Distribution-Free Two-Sample Tests for Scale”, Journal of the American Statistical Association 71 (353), 210-13.
› Hall, Angela, Neil Towers, and Duncan R. Shaw (2017), “Understanding how Millennial shoppers decide what to buy: Digitally connected unseen journeys”, International Journal of Retail &
Distribution Management, 45 (5), 498-517.
› Lemon, Katherine N. and Peter C. Verhoef (2016), “Understanding Customer Experience Throughout the Customer Journey.” Journal of Marketing, 80 (6), 69-96.
› Magidson, Jay and Jeroen K. Vermunt (2002), “Latent class models for clustering: A comparison with K-means”, Canadian Journal of Marketing Research, 20, 37-44.
› McCarthy, Daniel and Edward Wadsworth (2014), “Buy ‘Til You Die - A Walkthrough”, (accessed October 28, 2019) [available at https://cran.r-project.org/web/packages/BTYD/vignettes/BTYD-walkthrough.pdf].
VI. References
› Pantano, Eleonora and Viassone, Milena (2015), “Engaging consumers on new integrated multichannel retail settings: Challenges forretailers.” Journal of Retailing and Consumer Services, 25, 106-14.
› Peter S. Fader, Ka L. Lee, Bruce G.S. Hardie (2005), "RFM and CLV: Using Iso-Value Curves for Customer Base Analysis", Journal of Marketing Research. 42 (4), 415-30.
› Rohm, Andrew J. and Vanitha Swaminathan (2004), “A typology of online shoppers based on shopping motivations” Journal of Business Research, 57, 748-57.
› Vermunt, Jeroen K. and Jay Magidson (2002), “Latent Class Cluster Analysis,” in Applied Latent Class Analysis, 89–106.
› ——— and Jay Magidson (2004), “Latent class analysis”, The sage encyclopedia of social sciences research methods, 2, 549-53.
