Data profiling is a way to find out what is the profile of the information contained in the source. E.g. In a table a column may be defined as alphanumeric. However, majority of the data may be numeric. Profiling tools will provide the statistical information about how many records have pure no. populated as against no. of records with alphanumeric data.Before data migration exercise, these tools provide vital clues about whether the exercise is going to be a success or a failure. This can help is changing the target schema or applying cleanse at the source level so that most of the records can get in the destination database.In DW these tools are used at the design stage for the same purpose. Some tool vendors who sell this as a product call this as data discovery phase.