Sheetal Gandhi continues the series of exploring how different quantitative analysis techniques work and where they fit in. This month we look at segmentation analysis techniques.
Sheetal Padania 
17 Feb, 2009
Segmentation Analysis Techniques
In any given market, companies are unlikely to find that all customers react to sales and marketing activities in similar ways; instead it is likely that different groups of customers will be attracted to different products and services. As a result, many companies conduct segmentation studies to ensure that they target the right products and services at the right customers using appropriate sales and marketing activity.
In the 1980s Arcadia Group (previously Burton Group) conducted
market segmentation to ensure that each of its fashion chains targeted
different segments of the market rather than compete directly with one
another.
Market segmentation is a strategic tool that divides a market into subgroups
of customers. There are various ways that a market can be segmented but it
is now widely accepted that ‘needs based segmentation’ is the most useful way
because it allows companies to address their products and services to customers
who have common needs and therefore common reasons for choosing the product
or service
Segmenting customers can:
- Help create a stronger brand by delivering the right proposition to the key customers
- Strengthen customer relationships with the brand by being specific to their needs
- Identify unmet needs that extensions of the brand could exploit
- Help target strategies and activities to key customers
- Enable effective tracking of sales and marketing
Who to segment?
Segmentation studies within the healthcare market can be targeted at two key audiences: the health care professionals (HCPs) who make decisions about brand choice or the end users; patients and in some cases carers. There is also an opportunity to segment patients through the eyes of HCPs thus identifying how they segment their target audience. The choice depends on how decisions are made in a particular therapeutic area and what drives brand decisions.
When to segment?
Customers can be segmented anytime through the product lifecycle.
- Pre-launch – to understand needs, gaps and drivers in the market and to understand who the key players are
- Post-launch – to identify line extension opportunities, refine targeting and identify who are the current and future prescribers of the brand
What are the techniques used to perform segmentation?
Segmentation studies usually combine an initial qualitative stage to develop hypotheses and define basic parameters followed by a quantitative stage to validate and refine segments and provide the basis for targeting. As with many quantitative approaches, there are a number of different techniques available to conduct segmentation analysis, each with their own benefits and drawbacks. An overview of each of the most widely accepted quantitative techniques for segmenting a market is noted below:
Latent Class Analysis
Latent class analysis is one of the most sophisticated techniques used to segment a market. Compared with conventional techniques, it is able to identify segments with stronger differentiation, and can be used to uncover discrete groups that exist in complex markets, where choice is driven by both emotional and functional aspects.
Latent class analysis works by being able to predict cluster membership based on a set of measurements. It is a post-hoc method, and is therefore particularly useful for when clusters are unknown prior to the research taking place.
One of the main advantages of latent class analysis is that it produces robust statistics, which:
- Demonstrates how each variable contributes to the segmentation model
- Assesses the probability that every respondent belongs to every segment, thus providing a more accurate calculation of segment sizes compared to other methods.
Cluster Analysis
Cluster analysis is another popular classification technique for analysing
the market. It places respondents into distinct clusters by minimising the
differences between group members. Whilst conventional methods use demographic
data, cluster analysis relies on subjective elements or behavioural traits
to segment a market.
Cluster analysis is particularly popular due to being tried and tested and enabling fast turn-around market analysis. Nevertheless, other models have begun to replace this technique for a number of reasons:
- Cluster analysis cannot handle all types of data (such as categorical data)
- It excludes cases with missing values
- It is sensitive to outliers and to the inclusion of irrelevant variables
- It cannot predict membership of segments for new respondents
Discriminant Analysis
Discriminant analysis is also a valued tool for segmentation. It is used to build predictive models of group membership based on observed data. Unlike latent class and cluster analyses, it is a priori technique in that the segments are pre-defined before analysis begins. The analysis is, therefore, used to explain how each attribute best discriminates between the groups.
Discriminant analysis is particularly useful for confirming key drivers from existing segments, as well as predicting membership of segments for new respondents using ‘power questions’. However, it cannot be used to find initial segments in the market; and is an inappropriate
technique to use when there are high correlations in the data, as this can limit its effectiveness.
The following table provides a summary of the main analysis techniques available to segment a market, and the key characteristics of each:
| Technique | Method | Effectiveness for Segmentation | Effectiveness for Prediction | Statistical Properties |
| Latent Class | Post-hoc predictive | **** | **** | *** |
| Clustering | Post-hoc descriptive | **** | * | * |
Discriminant |
A priori-predictive | ** | *** | ** |
| Key: **** very good *** good ** moderate * poor | ||||
Summary
Within this article we have provided a short overview of why and when to conduct market segmentation, with whom, and which of the popular methods to use. Below summarises the four key points that should be remembered when conducting a segmentation study:
- ‘Needs based segmentation’ is the most useful way of analysing a market because enables products and services to be targeted to customers who have common needs.
- Market segmentation can be conducted at any time throughout the product lifecycle.
- Segmentation studies are best conducted with an initial qualitative stage to develop hypotheses, followed by a quantitative stage to validate the segments.
- Latent class is currently the most sophisticated technique available and should be used in most cases where segments are not pre-identified.
Your response:
Login to respond to this article: