Conceptually, OIS is a simple model.  A mixed integer and linear math programming (MILP) model, such as used for supply chain applications, is created of a portion of the firm’s income statement. It models the COGS and G&A costs of the income statement

Then, that model’s assumption of a fixed forecast is relaxed by inserting into it response functions derived from a non-linear math programming model such as used for marketing mix modeling applications. They model the remaining thew remaining portion of the income statement; the S of the income statement’s SG&A. Thus OIS is one, integrated model: OIS = bp (COGS+ SG&A).

Today, if the firm were to try to model its income statement with the same math programming techniques it would have to create two models: bp(COGS + G&A) + bp(S). However, bp (COGS+ SG&A) is necessarily superior to  bp(COGS + G&A) + bp(S). So, QED, OIS’s results are necessarily the best possible.