Having good COGS data is also helpful within a business to measure the profitability of different products: e.g. one brand vs another, or shirts vs pants. But i

Having good COGS data is also helpful within a business to measure the profitability of different products: e.g. one brand vs another, or shirts vs pants. But it's one of the hardest things to measure. Partly that's from the imprecision around the concept, partly from the complexity of the calculation (FIFO vs LIFO etc---the stuff the article said it was going to skip), and partly because it's not clear how to allocate some costs between products. And it can take a lot of manual data input work to track it all. There is a funny section of The Data Warehouse Toolkit by Ralph Kimball (the leading person behind the data warehouse consulting industry) where he says you should make sure to build the COGS data mart last, after you've already finished some easier projects and won the trust of the client, because it is the messiest and hardest, and also the most political. And when you do finally build it, v1 should be based on "rules of thumb". (Technically he advises building the profitability data mart last, but I think it's clear he's attributing the difficulties to the cost side, not the revenue side.) I once had a customer ask me to add COGS to their in-house ERP system, and I never even got started because they just couldn't answer simple questions about how they wanted it calculated.