Sheila Seles sent me the following link, which describes a new working paper by a Wharton professor and grad student.
They modify the methodology that we discussed in class today by having the set of "hit" and "niche" products change as the number of SKUs increases. Specifically, they define "niche" in percentage terms (e.g. the bottom x% of all SKUs) and find that when Netflix added a lot SKUs, the share of sales for "niche" products decreased.
The authors argue that this contradicts the "Long Tail" story. Chris Anderson says that it doesn't make sense to define "niche" in percentage terms and that the paper is flawed.
What do you think?