A good choice model should have enough scope to cover most of the factors that influence respondent choice. Our SUMM® model supports as many as 200 attribute-levels in 40 to 60 topic areas.
Many models handle only 20 to 100 attribute-levels. Users of more limited models are encouraged to focus only on "important" attributes. Assuming that "important attributes" can be identified a priori — something we have learned is not always the case — this concept remains problematic in two ways:
Studying fewer of the attributes that affect choice lowers the model's ability to accurately predict choice.
"Important" attributes are typically cost-of-entry in a category, and do not offer the capability to differentiate your products from competitors'. By including a larger list of second-tier attributes, your likelihood of locating unique and profitable strategies increases dramatically.
Many models handle only 20 to 100 attribute-levels. Users of more limited models are encouraged to focus only on "important" attributes. Assuming that "important attributes" can be identified a priori — something we have learned is not always the case — this concept remains problematic in two ways: