What is Master Data Management? MDM Explained for Beginners
One customer, three systems, three records that disagree. Master data management proves they're one person and builds the record everyone trusts. This is master data management explained from zero. It starts with what counts as master data (the nouns of the business) versus transactional data, and how nine rows end up describing four people. Then the three moves every MDM system makes: matching (deterministic rules, probabilistic scoring, and blocking, the way Splink does it), survivorship rules that build a golden record one field at a time, and a cross-reference table that maps every source ID to a master ID. We cover the four implementation styles (registry, consolidation, coexistence, centralized), run the whole process end to end on a worked example, compare commercial MDM suites with a warehouse-native "MDM lite" built on dbt and Splink, and clear up four common misconceptions. ⏱️ Chapters 0:00 - One customer, three records: which is true? 0:55 - What master data management means 1:29 - Master data vs transactional data 2:34 - How nine records end up as four people 3:44 - Move 1: deterministic and probabilistic matching 5:18 - Learned match weights and blocking 5:59 - Move 2: survivorship and the golden record 7:35 - Move 3: the cross-reference table 8:07 - The four MDM implementation styles 9:31 - Worked example: nine records to four 10:54 - MDM in the wild: suites vs MDM lite 11:50 - Four common MDM misconceptions 12:52 - Recap and limits 🔗 Links Splink: https://github.com/moj-analytical-services/splink dbt: https://www.getdbt.com DuckDB: https://duckdb.org Apache Spark: https://spark.apache.org Informatica: https://www.informatica.com Reltio: https://www.reltio.com Stibo Systems: https://www.stibosystems.com Profisee: https://profisee.com DAMA International (DMBOK): https://www.dama.org #masterdatamanagement #MDM #masterdata #goldenrecord #entityresolution #recordlinkage #datamatching #survivorship #Splink #dbt #datagovernance #dataquality #DuckDB #dataengineering