One of the major flaws of segmentation is the need for a 'simple segmentation'. Unfortunately, simple segmentations do not represent complex markets and complex human decision making. The averaging effect known as Simpson's Paradox means that the real differences between unique segments disappear when tucked into larger segments.
We build multi-tiered segmentations that roll up to a smaller number of larger segments, and then roll down to a higher number of smaller segments to reveal true differences between how consumers behave between segments. Our segmentation solutions are constructed so that all tiers are interconnected simultaneously, so it is easy to detect segment migration over time and to 'tune' the segments when needed.
Unlike traditional research, choice experiments/models capture real choices in real markets. They do this by asking consumers to choose their most preferred option from a set of alternatives, rather than to rate it. This overall preferred choice outcome more accurately reflects the way people make choices in complex markets. Structural equation models (SEMs) complement choice models because they reveal the factors that operate in a decision process.
For example, when looking for a laptop, the screen size and battery life are important to consumers, but we also need to know the factor order in the decision hierarchy: is it screen size first and then battery life as in the case of a creative designer, or is it batter life followed by screen size as in the case of a frequent flyer?