ATD methodology is embodied in our name: attention to data. We believe that the
key to success in data migration is to gain a real understanding of every item of data and its value to the organisation. Most importantly, we aim to establish the meaning of each data item as a piece of information rather than a string of letters and numbers.

The cornerstone of our approach is the meticulous accumulation of metadata (“data about data”). We do this through an expansion of the conventional data dictionary,
to incorporate all the information required to control and manage the migration process. Thus, in addition to the attributes that would normally be recorded against
a particular data item (data type, length, allowed values, etc), we would plan to capture the following information:

• rules for validating the contents of the data item
• an assessment of the relative importance of the data item, and the
  risk / impact of not migrating it successfully
• estimates of the time and resources required to achieve successful migration, including resources from I.T. and user departments
• tests to be carried out at every stage of the migration (e.g. to check the integrity of the source data, to verify any intermediary processing, or to
  check the data once migrated into the target system).

Having defined the necessary metadata, we can use the extended data dictionary
as the basis for generating the key deliverables of the migration project, including:

• specifications of the mapping between source and target data
• program code to effect the data migration
• test plans and test cases
• templates for reconciliation and sign-off of the migration
• project plans
• logs for monitoring issues and risks
• user-friendly means of viewing the source data, for cases such as
  migrations from external systems, where direct access to the source
  systems is not available.

ATD has developed a Toolkit to facilitate the generation of these deliverables.
Whilst we do not claim to have fully automated the data migration process,
we believe that our approach offers considerable savings compared to a traditional approach where data migration is treated as just another I.T. project. The same approach can be applied with equal effectiveness to other projects, such as data profiling or data cleansing, where a detailed understanding of the data and its role within the organisation is all-important.