Data Sciences Case Studies

Finding the right method

Identifying fundamental dimensions

Determining influences on the buying decision

A manufacturer’s product performs excellently in many respects – even better than its competitors. Despite this, its market share falls. Resources are often wasted because of a failure to identify the cause of this development. Products are often optimized on criteria that are relatively unimportant to the consumer.

The really important factors in the design of a company’s products or services can rarely be measured directly – e.g. by asking “How important to you is…?” Such queries only touch the cognitive level of decision-making, and can lead to serious misinterpretations and thus to costly errors in marketing decisions. The aspects that really govern the decision can be measured by means of procedures that take account of the underlying relationships. These include:

• Regression Analysis
• Discriminant Analysis
• Analysis of Variance
• CHAID Analyses
• Neuronal Networks
• Conjoint Analysis
• MaxDiff Scaling
• Structural Equation Models
• Penalty Reward Analysis PRA

Optimizing prices

What is the optimal price for a product? In today’s environment of globalization and market individualization, this question is becoming harder and harder to answer. Although it was once enough to simply add on an appropriate margin on top of production, sales and distribution costs to determine an effective market price, nowadays this approach is largely impractical. A strict orientation on the competition also fails to yield good results.

Yet price is a very important factor in almost all buying decisions. Concretely, price optimization may be required for determining the price of a new product, for price modifications in the event of product alterations (re-launches), or finding new price levels in order to achieve specific objectives such as e.g. boosting sales or optimizing profitability.

Alongside conjoint measurement, which allows for the analysis of an optimal price against a backdrop of product attributes and competitors, there are also methods whose focus is directed specifically towards price determination. These include:

• Conjoint Analysis
• Brand Price Trade Off (BPTO)
• Price Sensitivity Measurement (PSM)
• Gabor Granger

Identifying target groups

What are the expectations of my (potential) customers?
How can my employees be characterized?

Answers to such questions are important when decisions about strategies and activities in respect of technical feasibility and personnel issues need to be taken in the light of factors such as costs and capacities – in the fields of both marketing and human resources.

But they can rarely be answered in a global way, because very few people correspond to the “average”. Most people deviate more or less strongly from the overall norm. In consequence, in most cases it is not efficient to look simply at the average consumer or average employee. Nobody would feel that the message was really aimed at them.

Segmentation procedures make it possible to generate a multiplicity of sharp individual pictures from a diffuse overall picture. In most cases, the group as a whole can be broken down into 3 to 6 subgroups. All members of the total sample can be described relatively accurately by reference to this small number of subgroup profiles. Often, such profiles have a high recognition value for the expert observer. This underlines the practical relevance of such results.

• Cluster Analysis
• Discriminant Analysis
• Neuronal Networks
• CHAID Analysis

Market Positioning

The following situations frequently arise in market research practice:

• A consumer survey generates a vast number of ratings of the company’s own brand and those of the competitors;
• The results of an employee satisfaction survey can be broken down by department.

The presentation of such results in tabular form is often awkward, because it is difficult to recognize structures contained within such masses of numbers.

This complicates the central marketing task, which is to understand and strengthen the positioning of the company’s own offerings, and differentiate them from the competition, in a targeted and planned fashion. Only those companies or products with a clear and outstanding positioning in the market are well placed to compete.

Mapping procedures help in gaining a quick and clear overview – as in the motto “a picture is worth a thousand words” (or in this case “numbers”). They make it easy to recognize which products are similar, which products fill niches, and what their weaknesses are in comparison to the competition. Relevant procedures are …

• Multidimensional Scaling (MDS)
• Correspondence Analysis
• Mapping because of a Factor Analysis

Increasing loyalty

Customers and employees are a company’s most important assets. Keeping customers satisfied is often difficult, but it is generally much easier than winning new customers. Satisfied employees perform better than dissatisfied ones, and they are more loyal to the company – and thus more willing to go the extra mile in difficult situations. They transport a positive image of the company to the outside world.

These relationships are generally recognized. But measuring the key dimensions in a valid way and changing them in a positive and efficient manner requires special know-how and methods.

Customer and employee loyalty are complex latent variables, which can be quantified with the aid of various indicators. Often, the connections between loyalty and its influencing variables at different levels are equally complex. Loyalty again may have an impact on various aspects of behavior, such as the tendency to remain a customer of a company, to intensify the relationship, or to recommend the company to others – or the tendency to continue to work for a company over a long period and to support it with a high level of commitment.

Complex causal analysis procedures make it possible to quantify the direct and indirect influences in such interrelationships and to use these learnings in the context of CSM (Customer Satisfaction Management) and ESM (Employee Satisfaction Management) systems.