Customer Segmentation

From Open Risk Manual

Definition

Customer Segmentation (also Portfolio Segmentation) denotes the splitting of a Credit Portfolio (of any Financial Product portfolio more generally) into smaller components according to some defined characteristics.

Motivation

Customer segmentation is a core business Strategy irrespective of Risk Management considerations. In the context of Credit Portfolio Management it becomes an important tool toward identifying sub-sets of clients that exhibit similar risk characteristics. Segmentation can be either heuristic, based on existing labels / characteristics of the portfolio or statistical, based on historical performance, or, quite common, a combination of the two approaches.

Segmentation Dimensions

Heuristic Segmentation

The most basic segmentation of borrowers is into retail and corporate (and sovereigns) (reflecting the very different Credit Risk characteristics of physical persons versus legal entities).

Further differentiation and can be along dimensions such as:

  • the market segment (e.g. wealthy individuals versus regular borrowers in Consumer Finance, different business sectors in Corporate Lending)
  • origination channel (e.g. online, branch etc)
  • product segmentation (type of financial product)
  • geographic location
  • cohort (origination period or aged based segmentation)
  • relationship basis (existing / new customers)
  • operational details (e.g. portfolio size, origination system and data availability)

Statistical Segmentation

The above considerations can be denoted as heuristic, in that the segmentation is based on a-priori known categorizations. A portfolio can also be segmented on the basis of low-level characteristics and performance data using clustering techniques. Statistical segmentation may (in principle) produce segments that cut across heuristic boundaries

Issues and Challenges

  • Correct customer segmentation can be a key factor for successful implementation of a Credit Scoring model.

See Also

Contributors to this article

» Wiki admin