The first one evaluates the attributes. If the new product file contains new attributes respectively some previous attribute are missing, they were displayed in two separate tables (new attributes, missing attributes). You can check if these changes seems to be right. Otherwise you should abort the import process.
This step only appears if the data file is updated. As a consequence, during the initial uploading process of a new file this step is skipped, because new and old attributes cannot be compared.
