Disparate data is the root of all evil when it comes to attribution. Mapping data from various sources into a single source of the truth is necessary to establish a homogeneous data set for modeling. Don’t start modeling until your data is homogeneous.