> The only way that makes any sense is if you subtract returns for sales made in a different period to the sales period you are considering
Exactly. That's the way accounting works. They did not know in the previous quarter that the product would be returned in the following quarter, so they end up having negative sales in the current quarter.
Yes it produces "garbage output", which I find amusing.
There are multiple ways of calculating market share (e.g. units vs dollars or for different time periods) but assuming it is measured in dollars for a quarterly time period, how would you calculate the market share based upon my sample data above?
That is how it was calculated in a published trade magazine (either Infoworld or MacWeek, I think) I'm not sure if the the analysis was done by a market research firm or the magazine.
The story sounds fishy, but one way to have returns exceed sales is to ship 1,000 to stores on consignment, the store sells 500 of those for cash, the buyers return their 500 for refunds, and the store returns the remaining unsold 500. So 500 actual sales but 1,000 returns.
The 500 sent back by the consignment store aren't "returns" though, as they were never sold. That's just the equivalent of moving inventory from one warehouse to another.