Objective Driven Customer Segmentation

In the modern age of competitive buying and selling “KNOW YOUR CUSTOMER” has evolved as the primary and necessary focus area for any consumer driven Institution.

Customer Segmentation 

Customer segmentation is one such process that enables the business to segment the customers into multiple buckets significantly different in multivariate business relevant dimensions.

Machine Learning to the Rescue 

With the rise of the Data age, ML have touched upon and exposed hidden potential in every functional aspect of businesses.

Gone are the days when customers were segmented solely on the basis of uni variate analysis of demographic data. New age “Deep learning” ensembles with machine learning clustering algorithms have shown the state of the art benchmarks in customer segmentation as well.

Objective Driven Segmentation 

Objective driven segmentation comes to the rescue when the business has any objective wiz profitability of customers, sales etc and the idea is to understand the characteristics of major loosely coupled segments that vary in nature along such business objectives.

Sometimes the  ask of business is not to build a prediction model generating prediction scores for customers focusing on some business objective but instead understanding the nature of loosely coupled segments with varied affinity towards business objectives.

Taking this approach helps the business to understand the underlying multivariate characteristic of each segment and brushing up strategies specific to the segment.

Objective driven segmentation as a tool to derive the segments of customers with focus on specific business objective

Objective driven segmentation ensemble the prediction on objective and clustering approach to generate the segments varies on particular business objective.

Framework

Phase I Representation learning

This involves building a neural network trained on business objectives as outcome variables. Output of intermediate hidden layers of this network act as reduced representation of your customer data.

This representation is higher level concepts varied across the data in terms of business objectives.

Phase II Clustering 

This phase involves clustering the reduced representation obtained from the prediction phase. Clustering on this representation results in clusters with significant differences in terms of business objectives. 

Phase III Interpretation

To interpret the underlying nature clusters along dimensions of customer attributes . Cluster analysis is performed by mapping clusters with raw customer attributes. 

Statistical and descriptive study of clusters can enable the business to understand the nature of customer segments that vary across the direction of objective.

Conclusion 

The proposed framework has pushed off classical segmentation methodologies that suffer from paradoxes of categorical data , large no of customer dimensions and has proved to be state of the art general framework for customer segmentation.

This framework has the ability to be applicable across the business domains and segmentation scenarios given the business objective as direction of segments.