Con focuses on developing bespoke research methods and predictive models for human behavior, segmentation and develops text and data mining solutions for big data.
Con is a quantitative researcher and doctoral supervisor. He lectures in market research, managerial and marketing statistics, price modelling, decision modelling and segmentation, six sigma methodology and statistics, at both undergraduate and post-graduate levels.
His PhD is in measurement theory and applied market research methodology, with particular emphasis on the measurement of brand equity, using choice modelling and structural equation modelling to further develop an information economics model of brand equity, for the purposes of predicting market share.
His academic refereed journal publications and conference papers span branding, human machine learning, choice modelling, structural equation modelling, segmentation methods, research methodology, text mining, data mining, health care and psychopathy.
Co-Presenters: Christine Maddern
Market segmentation is the process of dividing consumers into sub-groups of consumers, which we call segments, based on some type of shared characteristics. A major weakness of market segmentation is how difficult it can be to decide which variables need to be used.
The usual criteria for segmentation include age, gender, region, ethnicity, income level, life-cycle position, buying habits, personality and motives and the client’s internal variables of interest. Focusing on one or several of these factors over the others can be the difference between success and failure, making market segmentation a risky pursuit.
Another weakness of market segmentation is that you might create groups that are too small to be profitable targets. For example, suppose a car dealership focuses on a narrowly defined income category, assuming this segment of the population is most likely to buy its cars. If that segment has too few members, the potential revenues will be too low to be worthwhile, no matter how successful the dealership is.
The same applies to most of our clients, where we believe that what makes sense to us will make sense in the market and that the market will respond.
The evidence is clear, most segmentations have a limited life span! We hear it time and time again, “we need to refresh the segmentation”, or worse still, “the market has changed and we need a new segmentation”. Sadly, the real reason has nothing to do with the market changing, except for cases where there is a major market change, and that is very rare indeed. It has all to do with the segmentation process.
We demonstrate that the problem with segmentation is with the choice of statistical models that are used to create the segmentation. We show 3 cases of the most popular segmentation models that are used in the majority of segmentation solutions and how they all get it wrong!
In our paper, we provide the leading robust way to segment customers, and a method that allows one to ‘tweak’ the segmentation as time passes, thus marking the segmentation future-proof!