In SAS® Enterprise Miner™, the Gradient Boosting node might complete without finding any splits.  In this case, the Number of Splitting Rules might be 0 for every variable, and Importance is 0 for every variable.  

There are no errors or warnings.  However, notes such as these are displayed in the node log.

A no-split scenario can occur when the Leaf Fraction property in the Gradient Boosting node is set to a value that is too high for your data.  

Use the Leaf Fraction property to specify the smallest number of training observations that a new branch can have.  The value is expressed as the proportion of the number N of available training observations in the data. N might be less than the total number of observations in the data set because observations with a missing target value are excluded. The default is 0.1.

One workaround for a no-split scenario is to specify a smaller value for the Leaf Fraction property.

Operating System and Release Information