Essentially, supply chain optimization boils down to two things:
1. Ensuring that all end-consumer demand is fulfilled (Implying higher Revenues from higher sales)
2. Filling this demand with the least amount of inventory in the pipeline (Implying lower costs due to less money tied up in inventories)
#1 - "Ensuring that all end-consumer demand is fulfilled" is not as simple as it seems. Typically most manufacturers don't have a good view of end consumer demand. They typically have good visibility till their distributors; post which haziness sets in. So visibility from distributors to retailers and from retailers to end-consumers is almost nil. And in economies where bulk of consumer sales happens via the small to medium format retailers (as opposed to Wal-mart type boxes where POS data is backward integrated to manufacturers), it can be a significant issue.
Most manufacturers however have a very good view of their Sales and as a result, often wrongly extrapolate this to Demand. The appropriate way to do this extrapolation would be to also track 'Stock-outs' at a SKU level, track promotion effectiveness and decompose their incremental impact on sales, etc. A fairly complicated area, filled by a few unique solutions - Oracle Demantra, SAP APO, etc. Have not used them, but they claim they accurately help a business manage Demand.
Another way of accurately measuring Demand would be to have unconstrained supply for a few months. So there are no chances of stock-outs and as a result Sales = Demand. Possible but expensive given that most manufacturers have hundreds of SKU's, and the costs to have unconstrained supply for a few months means a lot of working capital.
More on #2 - "Filling this demand with the least amount of inventory in the pipeline" in a subsequent post.