But in most cases, it would be difficult to determine the correlation between Products without conducting some sort of Conjoint Analysis study.
In many cases, and with some experience, it is easier to estimate the correlation between Features.
For example, objective Features, like megapixels and gigabytes, are highly correlated as most everybody would agree the more the better.
On the other hand, subjective Features, like brands and styling, are usually uncorrelated. Some Customers are fiercely loyal to a certain Brand, while other Customers are passionate about a competitive Brand.
In a Market Simulation designed to model a real-world Market, the user would start with best-guess Feature correlations and then set up the Market Simulation in a tuning loop.
This Building Blocks workflow examines Feature-level correlation. The workflow also introduces a new node, called the ‘Correlation Concatenation’ node, which allows groups of Features to be concatenated together.
But George still wants to borrow from his extensive industry experience to develop the best Product he can for the Market.
So instead, George breaks his Product down into a set of Features. He borrows from his Competitors to make his Product better, but he also works to ensure his Product is distinct.
Not that for the first time in the Building Block workflows, the Vertical Differentiation or importance of each Feature to Customers (the Mean and Standard Deviation) is specified directly in the Product Features table instead of in the ‘Matrix Distributions’ node. Functionality and Performance is more important (has a higher Mean) but provides less distinction than Style.
The Functionality of Jetson Gears is somewhat similar to Spacely Sprockets but not to Cogswell Cogs.
Functionality (input Matrix A) and Performance (input Matrix B) are correlated. But neither are correlated with Style (input Matrix C).
The ‘Correlation Concatenation’ node collects together the correlations of all Product-Features with all other Product-Features.
In this case, the Default Mean and Default SD (Standard Deviation) are ignored because these values have already been specified in the Input Attribute List.
The ‘Matrix Distributions’ node generates the Customer perceived values of each Feature.
The Product Correlation Matrix looks similar to the correlations from the previous workflow: BB-141 Three Competitors.
This Demand Curve for Jetson Gears is fatter, flatter, and more forgiving than the last Demand Curve.