CHAID, for Chi-square Automatic Interaction Detection, is a tree classification method useful for market segmentation. CHAID will "build" non-binary trees (i.e., trees where more than two branches can attach to a single root or node), based on an algorithm that is particularly well suited for the analysis of larger data-sets.

For example, it may yield a split on a variable Income, dividing income into 4 categories and groups of individuals in those categories that are different in some important consumer behavior (e.g., types of cars most likely to be purchased).

In market research, clustering types of customers can lead to very useful market segmentation. The correct classification of customers is essential for successful marketing strategies or product development. Customers can be classified on demographics or their needs (defined by usage, values, etc) by latent class.

For example, pharmaceutical companies need to classify the symptoms of mental diseases such as paranoia, schizophrenia for targeting at different groups of patients. Or telecommunication companies need to develop plans to satisfy the varying needs of different customers.

Conjoint analysis is used to determine how people value different features that make up an individual product or service. The objective of conjoint analysis is to determine what combination of attributes is most influential on respondent choice or decision making. A controlled set of potential products or services is shown to respondents and by analyzing how they make preferences between these products, the implicit valuation of the individual elements making up the product or service can be determined. These implicit valuations (utilities or part-worths) can be used to create market models that estimate market share, revenue and even profitability of new designs.

This technique is most frequently used to design new products and services, estimate price sensitivity, segment markets, measure brand equity, and develop competitive strategies. For example, a coffee maker manufacturer developing a new product wants to know the optimal price and the market share of the new coffee maker. A conjoint exercise would ask respondents to choose their products varying in prices, brand names, sizes, functions, and styles. By analyzing their preferences between products, respondents’ implicit values or utilities on the features and prices can be derived and these utilities can be used to determine optimal price and market share.

Discriminant analysis is to predict group membership based on a linear combination of the interval variables; the results from the procedure can give insight into the relationship between group membership and the variables used to predict group membership.

For example, a CRM manager may divide customers into two groups: loyal and disloyal customers. Discriminant function analysis could be used to predict loyal customers based on combinations of demographic variables and product/price preferences. The predictor variables might include age, gender, income, occupation, price sensitivity and product preferences, etc. The prediction model might provide insights into how each predictor individually and in combination predicted whether a customer will switch or not.

Factor analysis is used to reduce the number of variables, by combining two or more variables into a single factor. For example, satisfaction with a product or service, repeated purchase of the product or service, recommendation of the product or service to friends and family could be combined into a single factor such as overall satisfaction. Employee satisfaction can also be summarized in their satisfaction with pay, work environment, supervisors, coworkers, etc.

Latent Class (LC) models is a new tool to identify market segments in target marketing. LC models differ from the traditional modeling in the following ways:

• Do not rely on assumptions which are often violated in practice. Hence, they are less subject to biases associated with data not conforming to model assumptions.

• Can include variables of mixed scale types (nominal, ordinal, continuous and/or count variables).
• Demographic and other variables can be analyzed simultaneously for the identification of clusters.

There are three major kinds of LC models:

• LC Cluster Models will provide probability-based classification of segments. It is particularly useful for the development of a behavioral based and other segmentations of customers and prospects.

• LC Factor Models identify factors which group together variables sharing a common source of variation. Its marketing applications include the development of perceptual mapping which relate product and brand usage to behavioral and attitudinal measures.

• LC Regression Models classifies cases into segments and develops regression models simultaneously. They are useful to identify segments whether in mix of product attributes or in customer satisfaction.

Multiple regression, one of the most widely used research techniques, is to learn about the relationship between several independent variables and a dependent variable. It allows the researcher to ask the general question "what is the best predictor of ...". For example, a market researcher might want to know the best predictors of customer satisfaction or customer loyalty. If quality and price are used to predict customer satisfaction, the results will show that a certain amount of improvement in quality or lowering of price will lead to the fixed amount of improved customer satisfaction score. In a similar way, customer satisfaction score change can be used to forecast the increase of revenue or financial performance.

The real value of any study is the use of the results for decision making and we have experience with a wide range of optimization techniques that will help you arrive at the optimal decision and get the most value out of your analysis.

In addition to applying an optimal solution, optimization is a great test to the validity of your model. Nothing is more efficient than optimization when it comes to expose weaknesses in your model. The optimization will expose any weakness or inconsistencies in your model and substantially improve your understanding of how the model behaves.

Example of optimization:

Product line optimization: a discreet choice preference model optimizes your product line for market share, revenue or profit.

Perceptual mapping analyze and understand consumer perceptions of products. It produces a map of a market to show how products are perceived on specific features or attributes such as reputation, price, quality, etc.

Perceptual maps show which products compete in the consumer's mind and suggests how a product can be positioned to maximize preference and sales. They provide valuable insights for a number of marketing decisions. Some major applications include:

Identifying Product Weaknesses

Identifying Differences Among Groups

Structural equation models (SEMs) describe relationships between variables. They are similar to combining multiple regression and factor analysis. SEMs also offer some important, additional benefits over these techniques including an effective way to deal with multicollinearity, and methods for taking into account the unreliability of consumer response data.

In addition to examine cause and effect relationship, SEMs can measure and model not directly observable variables such as brand attitudes, customer satisfaction, perceived value, repurchase intentions and perceived quality. The modeling results provide valuable insights as to what an organization can do to most improve quality, perceived values and customer satisfaction.

It was originally based on the needs of media schedulers to maximize reach and frequency of media spending across different vehicles (print, broadcast, etc.). In a research context, TURF provides estimates of market potential (i.e., number of users reached and/or their frequency of usage) – typically in the context of a line configuration problem. For example, there may be 10 possible flavors for a new yogurt, but the retail trade will only take three. The TURF algorithm identifies the optimal product line to maximize the total number of consumers who will purchase at least one SKU and, at the same time, minimize consumer overlap across all the flavors. TURF provides marketing managers with answers to three questions:

It was created by Dutch economist Peter H. van Westendorp to assess consumers’ price perception. It is based on the premise that there is a range of prices bounded by a maximum that a consumer is prepared to spend and a minimum below which credibility is in doubt. It is a price sensitivity meter based on respondents’ answers to four price-related questions:

The responses to the above four questions will be plotted and the key intersections on the curves can be used to interpret price perceptions.