Large population skewing Standard Deviation?

lbour

New Member
My company makes 3 products (X, Y, Z) and there are 4,000 sellers of these products. Company A has asked me if their sales mix is in line to that of their peers. Hence, I'm trying to compare Company A's sales mix to that of the entire market. For Co A, I calculated Revenue from X/Total Rev to get X's % of contribution to the total... and did same for Y & Z. I summed X across all 4,000 sellers and did the same calculation to get the average Revenue % mix for X. I want to compare the % mix of X for Co A to the % mix of X for the economy to answer whether Co A's mix of X is relatively higher or lower than that of their peer set. This gets tricky because Revenues are confidential information and the % mix of each product (X, Y, Z) for the Market is confidential. In short, I can't tell Co A what the % of mix for X is for the total market. I am thinking though that I could calculate how many standard deviations their sales mix for X is from that of the average of the 4,000 sellers. But what I'm finding is that, say, 500 sellers sell a lot of X... and the other 3,500 are selling 1 or 2 units. Hence the mean and standard deviation is really small... so comparing to Co A's % mix doesn't give me useful information. Is there a valid way to eliminate the impact of the long tail? Or is my idea of trying to measure how many standard deviations Co A's % mix is from the average of the other sellers faulty?