What is Data Mapping?

Any integration between two systems runs into this immediately: one CRM calls a field "full_name" while another splits it into "first_name" and "last_name," and a mapping layer has to define that translation, including any format conversion (date formats, currency units, enumerated value lists) needed along the way. Get the mapping wrong and data doesn't just fail to transfer - it transfers incorrectly, which is often worse because it looks like it worked. Mapping gets harder as schemas evolve; a field that used to map cleanly can silently break when the source system renames or restructures it, so mature integrations version their mappings and add validation that catches a broken mapping before bad data flows downstream rather than after. AI has started to help with the initial mapping-design step itself - inferring likely field correspondences from sample data and names - though the final mapping still typically needs human confirmation, especially for fields with business-specific meaning.