Segmentation models identify discrete groups in which there are higher percentages of desired individuals.
For example, they can be used to differentiate between responders and non-responders to a direct mailing campaign, or to compare the characteristics of high and low value customers.
Segmentation models can be built using many different variable types:
- transactional variables (e.g. recency, frequency and value of purchases)
- geodemographic variables
- sociodemographic variables (e.g. age, income)
- lifestyle and attitudinal data (e.g. hobbies, newspaper readership, product holdings)
They can help you to develop and position products, by providing an increased understanding of the composition of markets and your customer base, and determine the creative tone and content of your marketing activities.
We use CHAID and Cluster Analysis to develop segmentation models.
CHAID is relatively simple for a non-statistician to understand and is highly visual. It will build non-binary trees (i.e., trees where more than two branches can attach to a single root or node), based on a simple algorithm that is particularly well suited for the analysis of larger datasets often found in marketing. It is also very useful in identifying interactions between variables and thereby enhancing propensity models.
Cluster analysis aims to allocate individuals to a set of mutually exclusive, exhaustive groups, so that individuals within a group are similar to one another, whilst individuals in different groups are dissimilar.
Cluster analysis can be employed for a variety of objectives including:
- data exploration
- data reduction
- hypothesis generation
- prediction based on groups
We can develop cluster models to describe:
- customer behaviour – often an essential first step as an organisation evolves from a product or channel focus towards a customer focused business
- customer potential or share of wallet – a good starting point for developing a marketing plan
- values and attitudes – to fine-tune brand positioning and products