Save the Children wanted to investigate the potential of data science to improve their ability to segment donors successfully. These segments can then be targeted with specially crafted marketing campaigns to increase overall donations. Previously, segmentation was done using predefined demographics but it is not clear whether age, gender and other demographics provided the best segmentation’s.
A clustering algorithm was used to identify the most natural grouping of donors from past donation behaviour such as frequency and value of donations. All data points were anonymised. Four natural groupings were identified and by analysing common key factors in each group (profiling), we were able to train a neural network to predict which new potential donors belong to which group
Identifying natural key metrics allowed for more accurate segmentation groups. More tailored content could be developed and distributed through the right media channels. A 2 day training course was also delivered and the next stage is to refine the process and scale up.