Ideally, you will find correlations between the variables and then you could do a factor-analytic compaction of the variables. However, there are at least two difficulties that need to be considered in this context: The variables mentioned can develop in different ways, that is to say that a company may achieve high revenue growth through certain measures, but not an improved return on equity (for example because of a rise of its capital). Then there would be no (positive) correlation. And further, one must bear in mind that the variables margin growth, revenue growth, and return on equity are not necessarily valid indicators of organizational learning, which can also cause many other effects. The variables mentioned are also based on other reasons (i.e. financing strategy, business strategy), which are maybe not covered in your design. It seems to me more favorable to include internal variables as outcome of organizational learning.

Apart from the inedequencies/difficulties described by Axel Faix, u may use the average of the three variables as rough estimate