Horizon CDT Research Highlights

Research Highlights

Using Transaction Data To Understand Consumer Behaviour During Life Events

  William Darler (2013 cohort)   williamdarler.wordpress.com

Research Aims

Life course studies describe how events experienced by people over the course of their lives can lead to changes in behaviour over time [1][6]. Empirical studies in the marketing domain have investigated changes in consumption behaviour by using data gathered from surveys [1] [2][4] or interviews [3][7] and there has been little longitudinal research that uses transaction data to see how shopping patterns change over time during a life event.

Our research has two main aims (1) Predict the times at which changes in consumption behaviour take place, and (2) Segment customers by their behavioural trajectories over time. We draw together concepts and techniques from multiple disciplines to show how big transaction data sets can be used to examine customer behaviour during life events.

Method and Data

We create frameworks to describe how groups of customers experiencing an event can be selected from transaction data, how the event can be divided into different stages based on their consumption patterns, and how to predict which customers are experiencing each event-stage.

Customer selection

Our initial sample was a transaction data set of 3.4 million customers from a multinational health and beauty retailer. These customers bought around 650 million items from 455 product categories over a 10-year time period between 2006 and 2015. From this data set, we selected subsets of customers who experienced three different events: Parenthood, Flooding, and Christmas. We chose these events because they affected different populations of consumer. Parenthood is a personal event that affects an individual or family unit, Flooding is a local event that affects a community, and Christmas is a national event that affects a whole country.

Timing and segmentation

To identify the times at which consumption behaviour changed for each of our three events, we used an e-divisive algorithm from Matteson & James' [5] ECP package for R. The algorithm detected the points at which significant changes occurred in our multivariate time series data set. To segment our customers by their behavioural trajectories, we used a dynamic topic modelling algorithm in Python. The algorithm detected changes in the groups of items that appear together in customer baskets over time, clustering together customers who made similar purchases over the course of each event.

Key Contributions

This study contributes to analytical CRM, data mining and life course literature by providing examples of how a large transaction data set can be used to identify the times at which consumption behaviour changes during different life events, and segment customers based on their behavioural trajectories. It provides the foundations for novel frameworks that could be used by managers to target customer segments for direct marketing material or product recommendations at appropriate times during life events that they may be experiencing. It could also be useful to help managers merchandise, and plan stock or staffing levels depending on the population experiencing these events and the location of the events themselves.


  1. Andreasen, A. R. (1984). Life status changes and changes in consumer preferences and satisfaction. Journal of Consumer Research, 11(3), 784-794.
  2. Du, R. Y., & Kamakura, W. A. (2008). Where did all that money go? Understanding how consumers allocate their consumption budget. Journal of Marketing, 72(6), 109-131.
  3. Karanika, K., & Hogg, M. K. (2013). Trajectories across the lifespan of possession-self relationships. Journal of Business Research, 66(7), 910-916.
  4. Mathur, A., Moschis, G.P. & Lee, E. (2008). A longitudinal study of the effects of life status changes on changes in consumer preferences. Journal of the Academy of Marketing Science, 36, 234-236.
  5. Matteson, D. S., & James, N. A. (2014). A nonparametric approach for multiple change point analysis of multivariate data. Journal of the American Statistical Association, 109(505), 334-345.
  6. Moschis, G. P. (2007a). Life course perspectives on consumer behavior. Journal of the Academy of Marketing Science, 35(2), 295-307.
  7. Rossi, P. E., McCulloch, R. E., & Allenby, G. M. (1996). The value of purchase history data in target marketing. Marketing Science, 15(4), 321-340.


Darler, W., Goulding, J., Smith, A., & Roberts, D. (2017) A framework to segment life events using customer transaction data. 2017 Winter American MArketing Association Conference (AMA), Orlando, Florida.

Roberts, D.L., & Darler, W (2017). Consumer co-creation: an opportunity to humanise the new product development process. International Journal of Market Research, 59(1), 13-33.

This author is supported by the Horizon Centre for Doctoral Training at the University of Nottingham (RCUK Grant No. EP/G037574/1), by the RCUK’s Horizon Digital Economy Research Institute (RCUK Grant No. EP/G065802/1), and by N/A.