September 14, 2017

September 14, 2017

September 14, 2017

September 14, 2017

May 26, 2017

Please reload

Recent Posts

Data Storytelling

May 26, 2017

1/2
Please reload

Featured Posts

Tandem segmentation: Getting your product to sell itself

A case study application of how a fictional food maker could use tandem segmentation to better understand customers.

 

Segmentation is the process of classifying a market or customer base into distinct attitudes or behaviors. The segmentation process entails segment identification, segment characterization, evaluation and target segment selection. Market structure, consumer perceptions, shopping behavior and branding images are all taken into account.

 

There are many ways to make better decisions and improve market ROI. Segmentation can help the client: enhance new product development; make sense of patterns of consumer behavior linked to a brand or product; and determine the motivations for consumers to buy their product or service.

 

Segmentation cuts in different ways (pun intended!). We are going to focus on one. I will describe a particularly effective tandem method we often deploy for comprehensive segmentations. The tandem method is to conduct a factor analysis, followed by a cluster analysis. The technique is post hoc (data collected from a consumer survey) and covers not only the usual suspects of frequency and purchase behavior but also lifestyle and attitude issues that are associated with product usage. It combines many different dimensions of brand consumption and blends them into specific and informative characterizations.

 

Our fictional client is the Guaranteed Food Corporation (GFC). GFC is commissioning a major study for its Health Valley line of products, which includes healthy selections for breakfast, granola bars and other processed healthy food choices. The sample is comprised of those respondents who claim that they purchase healthy food alternatives on a regular basis.

 

The lengthy questionnaire explored a range of purchasing, behavioral, health and lifestyle attitudes. GFC’s goal was not simply to identify its core customers but to also understand them. The company wanted to continue its bond with Health Valley’s main consumers and also attract more like-minded people to the Health Valley product line. GFC wanted to dig deep to give the Health Valley brand team the full picture. We were going to provide it.

 

Measure of separation

 

Every post hoc segmentation needs a measure of separation between the final segments. These are often referred to as business rules. The three most common business rules are: latency of purchase, frequency of purchase and amount spent.

 

Consumption variables in the Health Valley study are de facto business rules. They are variables within the study that allow us to compare segments based on important discriminators for the Health Valley brand. Consumption variables are descriptive; that is, they are used for comparison purposes between segments and are not input into the cluster analysis itself. Below are the consumption variables – business rules – that we used to compare Health Valley segments:

  • monthly spending on packaged health food;

  • percentage of segment that are Health Valley customers;

  • number of Health Valley products purchased in past month (non-customers=0).

Among other questions, the Health Valley questionnaire contained arrays of attitudes that respondents were asked to rate on a discrete (1-to-7) scale. For the segmentation, the following dimensions were queried:

  • food purchase behaviors;

  • exercise and wellness activities;

  • lifestyle attitudes;

  • shopping behavior;

  • health food attitudes; and

  • motivational reasons for healthy living.

The first step is to use a common marketing research technique, principal components analysis, commonly referred to as factor analysis. Factor analysis finds underlying structures of association between variables. Put another way, factor analysis creates “families” of attitudes that tend to be rated similarly. Figure 1 shows one example output of the six factor analyses run for this study. The names of the factors at the top (e.g., Healthy Food Shopper) are subjective and are normally based on the attitudes contained in the analysis. Figure 2 shows the consumption variables by each of the three shopping behavior segments.

 

 

 

When programmed, factor analysis creates variables for each factor. Each respondent receives a factor loading (similar to a correlation coefficient) for each factor. For our example, three new variables were created. Each respondent received a score for each of the new variables. A respondent is placed into the family where he/she has the highest score.

 

For the six Health Valley factor analyses we set each factor analysis to have three factors (they generally run between three to six). The table in Figure 3 summarizes the families (factor analysis results) that were created.

 

To sum up, what we have created are six new variables, each with three values. Each respondent has one value in each of the six new variables.

 

Perfect second step

 

K-means cluster analysis aims to partition observations into x number of clusters in which each observation belongs to the cluster with the nearest mean, serving as a framework of the cluster. K-means is not my favorite clustering algorithm: if the data is varied, the results can be murky. The k-means algorithm, on its own, is too sensitive to outliers. However, for our tandem method, it is the perfect second step. I’ll explain below.

.

.

.

 

Vital marketing tool

 

Not every dollar spent on advertising and sales is created equal – some of those dollars generate far more revenue than others. Segmentation is a vital marketing tool. Peter Drucker had it right when he spoke of the art of segmentation and its marketing extension: “The aim of marketing is to know and understand the customer so well the product or service fits him and sells itself.”

more...

Share on Facebook
Share on Twitter
Please reload

Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square

© 2015 by Savvy Analytics & Savvy Web Design and Development