I am at design stage and was wondering if CBC HB can support hybrid rolled-up estimation. Below are the specs:
There are 36+ products + new product X in the market so typical CBC would not work.
Instead, I am showing product file in traditional way and asking following questions:
Q1. Purchase likelihood (Rating scale)
Q2. Likelihood to replace product currently use (PIPED from stated question before conjoint - 36 products).
Q3. Likelihood to replace product with key competitor (only one) [SHOW IF AWARE OF competitor brand]
I will be showing Q2 and Q3 in traditional conjoint format where scenario 1 will be simulated features showed in Q1, scenario 2 will be Q2 (no levels just brand name) and scenario 3 will be Q3 if applicable.
My question is:
1. Does this approach makes sense.
2. For analysis level, I will rollup the brands by manufacturer to create manufacturer portfolio (approximately 3-4 manufacturers). Essentially, we are not interested in brand level but only at portfolio level and cannibalization effect. How would I estimate utilities for this type of setup?
This is interesting yet complex approach and never tried before. Any input will greatly help to pull the strings together.