In today’s competitive and Omni-channel markets, the key to success of each business is understanding its customers. Through a systematic approach of harnessing data, leveraging analytics and knowledge, one can drive effective marketing decisions in a technology-enabled and model-supported interactive decision process.
We live in a diverse society where customers differ in their values, needs, wants, constraints, beliefs, and incentives to act. To increase the chance of creating customer success, products (or services) need to be completive and differentiated to satisfy the needs of customers. Marketing engineering is defined as “a systematic approach to harness data and knowledge to drive effective marketing decision making an implementation through a technology-enabled and model-supported decision process”. Marketing Engineering segments groups of customers with similar wants, needs, and responses; targeting then determines which groups should be best served; and positioning elevates stature of the offering in the targeted segment versus the competition.
Through the segmentation process, we get a better understanding of the prospective buyer and align our marketing efforts efficiently. We target an optimal segment—not to rely on a on- size-fits-all common marketing program, nor developing a unique and costly non repeatable approach for each customer. This establishes a perspective and personas where members are different between segments but similar within personas in each segment. This is an extremely powerful and more scalable approach , because with sufficient differentiation comes perceived value and the sales opportunity!
Data and Analytics to the rescue
To illustrate the segmentation process, we use an example of a leading brewer, launching a new beer product in a crowded and competitive market. How should we segment beer consumers, with diverse taste and drinking habits?
We obtain results of a market survey of hundreds of responders’ preferences, likes and drinking habits. They are represented by the attributes, e.g.: rich full-bodied taste, refreshing, for drinking with friends, gives a “buzz” etc.—each ranked 1 to 9.
Armed with that data, we will use the cluster analytical function in Tableau to partition the responses into a few segments. Clustering takes all the preference variables as an input and groups the respondents into similar segments. In this case, we visualized the individual respondents as colored marks and overlaid the resulting clusters, where marks within each cluster are more like one another than they are to marks in other clusters. It is a business decision as to how define and choose how many segments to use. In this example, after a few iterations, we decided on three.
Data Science is the Secret Sauce
Under the hood, Tableau uses k-means clustering that splits data into K segments, locates their mean based centers, and minimizes distances between the individual attribute points. The k-means are determined by the algorithm with squared Euclidean method for each k. Even though this sounds complicated, this is basically the Pythagorean theorem, calculating the distance between two points on the X and Y axis—in this case a bit more complex, as we have more attributes.
Among other metrics, Tableau calculates the statistical significance, the p-value for each attribute, helping the analyst gain the confidence and assess the model’s accuracy before recommending the segmentation structure.
The segmentation results should be then repeated with the discriminant analysis of the respondent’s demographics (e.g. age, income, status), to increase the accuracy and effectiveness of target marketing programs.
Segmentation helps address a core strategic issue facing all companies: Which markets or customer groups should we target to serve? How do we align our product offering to customers’ needs in those segments? How do we rationalize marketing spend and make it segment specific? Used with the right data and supported by modern analytical tools, segmentation is an extremely powerful tool and in the hands of any marketing engineer.
In our next blog, we will demonstrate how to curate data with self-service preparation and integration techniques.