Model training includes estimating and optimizing saturation when you tag a variable for saturation optimization during the variable selection step. Model training results then includes saturation estimates such as this:
The saturation curve shows the diminishing impact of a given independent variable on the dependent variable. In this example, the effect of media spend on Banners reaches its plateau at about 500 and is largely ineffective past that mark.
If the variable is found to have no saturation effect, its saturation curve will look as follows:
In this case the independent variable shows little to no diminishing impact: as the independent variable increase (see 'TV' on x axis), so does its impact on the dependent variable.