Additional valuable properties of the IPM

In most product optimization procedures which exphcitly estimate ideal points (LSA, EDIPM, BaSic, etc.), it is assumed that each consumer is associating the product set to one unique ideal product. Although this assumption is reasonable in the case where the product set consists of very similar prodncts, it is not necessarily correct when the products are more heterogeneous. Let us consider a test in which milk chocolates and dark chocolates are compared. In such case, it is fair to say that consumers might like both the milk and the dark chocolates, and hence have one ideal for the milk chocolates and one ideal for the dark chocolate. The optimization procedure should then associate the milk chocolates with the ideal of reference defined for the milk chocolates, and the dark chocolates with the ideal of reference defined for the dark chocolates. 

The decision whether one or multiple ideals should be considered is a decision that should not only be based on statistical grounds, but also on the business arguments. If two or more latent subcategories exist, is it worth formulating more than one optimal product Or should only one optimal product be formulated Such a decision is obviously based on the size of the differences between the ideal of each latent subcategory, and on a business decision as to whether one or multiple products should be placed on the market. 

As mentioned in Section 14.2.3, the validity of the ideal data needs to be checked, since the data are obtained from consumers who are asked to rate a Active product. For that reason, before any optimization, one has to make sure that the consumers are providing relevant information when rating their ideals. 

Since the ideal product has both a sensory and a hedonic dimension - the ideal being defined as a product with particular sensory characteristics that would maximize liking - checking for the consistency of the ideal data should be done both from the sensory and the hedonic point of view. These consistencies are checked both at the consumer and at the panel level. 

Rapid Sensory Profiling Techniques and Related Methods 

---