Lesson 1 — Shared Context and the Four Data Types
What Chapter 10 is solving
Large organizations rarely have only one representation of the real world. A customer can appear in CRM as C-992, in billing as 77410, and in a support application under an email address. Meanwhile one application stores country as US, another as USA, and another as 840.
Those examples look similar because both involve inconsistency across systems, but they are different management problems.
- “Are these two customer records the same real person or company?” is a Master Data identity question.
- “How do these three country codes represent the same controlled concept?” is a Reference Data mapping question.
That distinction is the doorway into the chapter.
The four-way comparison
Reference Data — controlled choices and classifications
Reference Data contains managed values used to characterize, classify, constrain, or translate other information. Examples include order status, country code, industry code, product category, risk class, or a mapping between code standards.
The deciding question is: What allowed or standardized value describes something else?
Master Data — persistent business entities
Master Data identifies and describes core entities that the enterprise repeatedly reuses. Examples include a particular Customer, Product, Location, Supplier, Legal Entity, or Account.
The deciding question is: Which real-world thing is this?
Master Data often needs identity resolution because multiple records may refer to one entity—or superficially similar records may actually represent different entities.
Transaction Data — the event
Transaction Data records business activity: an order, payment, shipment, invoice line, claim, service event, or another occurrence.
Transactions typically involve Master entities and use Reference values. An invoice might involve Customer 84721 and Product 510 and use status Paid.
The deciding question is: What happened?
Metadata — information about the data
Metadata describes or manages other data: definitions, source system, steward, version, effective date, update frequency, lineage, format, or quality rule.
For a country-code Reference set, US and USA are values. “Maintained by ISO; steward = Enterprise Data Office; refreshed annually” is Metadata about that set.
The deciding question is: What tells me what this data means, where it came from, or how it is managed?
Shared-context stack
A useful mental picture is:
TRANSACTION EVENTS → involve MASTER ENTITIES → classified/contextualized by REFERENCE DOMAINS
METADATA → describes meaning, authority, stewardship, lineage, and lifecycle across all three.
Do not turn Metadata into another business entity domain. It spans and describes the other data.
Why “shared” does not mean “Master”
A frequent trap is to assume that anything used enterprise-wide is Master Data. An enterprise status list can be highly shared and important while still being Reference Data. Conversely, a customer entity is Master Data even if only two systems currently use it.
The classification follows the role the data plays, not merely how important, centralized, or widely used it is.
Three program goals
Chapter 10’s management intent is to:
- create complete, consistent, current, authoritative shared Reference and Master Data;
- enable enterprise sharing and reuse; and
- reduce risks and integration costs caused by inconsistent definitions, identifiers, values, and semantics.
Guiding governance ideas
Shared data belongs to the enterprise rather than to one application. It needs defined authority, quality monitoring, stewardship, controlled change, and accountability for enterprise use.
Example
A hospital has two records:
- Patient ID
P-1088, name “Maria J. Soto” - Patient ID
887004, name “Maria Soto,” same DOB and phone
The correct first classification is Master Data/entity identity, not a code-standard problem.
If the same hospital stores visit status as C, Closed, and 3, then the problem becomes Reference Data mapping/standardization.
Stop and check
A data dictionary says the Customer_Status domain is owned by Sales, refreshed monthly, and must use values Active, Dormant, Closed.
- Active/Dormant/Closed = Reference Data.
- Owner, refresh schedule, and definition = Metadata.
- Customer 84721 = Master Data.
- Customer 84721 places Order 3301 = Transaction Data.
Source anchor: DAMA-DMBOK2 Revised, Chapter 10, pp. 329–340.