Joining the dots

22 Feb, 2013

In at the data deep end.

Where to start? Well, the beginning is usually a good place, so the first area for exploring patterns in data is the audience segmentation and profile building process. Crimtan uses interest and intent behaviour demonstrated online to define how different profiling attributes are related in order to create patterns and groups among users. Identifying patterns in data and implementing data mining models could provide a better understanding of why certain segments of campaigns may be performing poorly or well, or predict in which diverse range of categories a customer may be interested. In other words, extracting knowledge of data regarding segments could provide intelligent decision-making outcomes.

The challenge at the segmentation process is to drive ad placements efficiently – that means with confidence and at scale across different audience categories. It is commonly known that different customer types visit the same website for different purposes. It is essential to identify groups of similar customers, namely segments, in order to enhance the company’s marketing plan. The more you know about these segments the better the audience targeting will be. Consequently data mining models are developed with the aim of identify affinities between segments that could lead to the augmentation of campaign’s performance.

With the aim of exploring patterns, Crimtan’s behavioural data is used in order to discover correlations between user’s segments. These correlations correspond to groupings of segments and these groups can be investigated in more detail and profiled by modelling.

In particular data mining is used in order to

  • Identify groups of users based on the types of segments they belong

But this is not enough….

  • Through data mining analysis we will reveal what characteristics these users appear to have

One fundamental phase of data mining is that the creation of a model is an iterative process. The results of data mining trigger new questions which in turn can be used to target the most profitable segments by refining and improving the characteristics of the segments.

But back to work for me. More insights next time.

Maria