Master Data Management Flashcards

Question 1

What is master data in an organization?

Accepted Answer

Master data is the core, slowly changing information that describes the key entities a business relies on, such as customers, products, suppliers, employees, and locations. It gives every system a shared, agreed description of those things, so that a customer in billing and a customer in shipping refer to the same real-world party.

Question 2

How does master data differ from transactional data?

Accepted Answer

Transactional data records events that happen at a specific time, such as an order placed, an invoice paid, or a shipment sent. Master data describes the people, products, and places those events involve. Orders change constantly, while the customer placing each order changes rarely.

Question 3

What is reference data, and how does it relate to master data?

Accepted Answer

Reference data is a controlled set of codes and categories used to classify other data, such as country codes or currency codes. Master data often points to it, for example when a customer record stores a country code from a standard list. Reference data is usually simpler and changes less often than core business entities.

Question 4

Why do organizations need a single trusted version of their master data?

Accepted Answer

Organizations need a single trusted version because the same entity often appears in many systems with different names, spellings, and identifiers. When sales, finance, and support each hold a different customer record, reports disagree, campaigns reach people twice, and decisions rest on conflicting facts. A golden record gives everyone one reference to build on.

Question 5

What is a golden record in master data management?

Accepted Answer

A golden record is the single best version of an entity, built by combining the most reliable attributes from several source records that describe the same real-world thing. It is assembled through matching, cleansing, and survivorship rules, and it becomes the reference that downstream systems use.

Question 6

What is data stewardship, and why does it matter for master data?

Accepted Answer

Data stewardship is the assigned responsibility for the quality, definition, and appropriate use of a set of data. A data steward resolves disputed records, approves changes to standards, and makes sure business meaning stays consistent. Without named stewards, master data drifts, because no one is accountable when errors appear or when systems disagree.

Question 7

What does data governance cover in an MDM program?

Accepted Answer

Data governance is the set of policies, roles, decision rights, and processes that determine how data is defined, owned, accessed, and corrected. In an MDM program, governance sets who may create or change a customer or product record, which standards apply, and how disputes are escalated. Technology enforces these rules, but governance decides them.

Question 8

Which roles typically take part in an MDM program?

Accepted Answer

Typical roles include an executive sponsor who funds and champions the program, a data owner who is accountable for a domain such as customers, data stewards who manage daily quality, and an MDM architect or administrator who configures the platform and integrations. Business analysts and data quality specialists often support them.

Question 9

What are the main implementation styles used in master data management?

Accepted Answer

The commonly described styles are the registry, consolidation, coexistence, and centralized hub. They differ in where the master record is stored and how much control the MDM system has over source systems. Organizations choose a style based on how many systems share data, how mature their governance is, and how urgently they need consistency.

Question 10

How does a registry style of MDM work in practice?

Accepted Answer

A registry style leaves master data in source systems and stores only an index of identifiers that links matching records. When a user asks for a complete customer view, the registry tells the system which source records belong together. Source systems stay unchanged, making this the least invasive style.

Question 11

How does a consolidation style MDM differ from a registry?

Accepted Answer

A consolidation style copies data from several source systems into a central hub, where matching and survivorship rules build a golden record for analysis. Source systems keep their own data and are not updated by the hub. This suits reporting well, but source changes do not flow back automatically.

Question 12

What is a coexistence style, and when is it useful?

Accepted Answer

A coexistence style maintains a central master record that is synchronized in both directions with source systems, so updates made in either place are propagated to the other. It is useful when several systems must keep working as they do today while the organization gradually moves toward consistent data.

Question 13

What is a centralized hub style, and what trade-off does it involve?

Accepted Answer

A centralized hub style creates a master data system that owns the authoritative record, and source applications read from and write to the hub rather than maintaining their own copies. This gives the strongest consistency and governance. The trade-off is greater disruption, because applications must be re-integrated, and the hub becomes a critical dependency.

Question 14

What does a survivorship rule determine when merging duplicate records?

Accepted Answer

A survivorship rule decides which value from competing source records is kept in the golden record for each attribute. For example, the rule might keep the mailing address from the most recently verified source or the phone number from the system with the highest data quality score. Rules are usually set per attribute.

Question 15

What is the purpose of matching in master data management?

Accepted Answer

Matching identifies which source records refer to the same real-world entity, even when names, addresses, or identifiers differ slightly. Matching typically compares attributes such as name, address, date of birth, or tax identifier, and assigns a score that indicates how likely two records are the same.

Question 16

Why is it important to test match rules against real data samples before deploying them?

Accepted Answer

Testing matters because match rules that look sound on paper can produce false positives, merging unrelated customers, or false negatives, leaving duplicates in place. Testing against representative samples reveals these errors and lets teams tune thresholds and attributes before bad links spread through downstream systems.

Question 17

What is data quality, and which dimensions are commonly used to measure it?

Accepted Answer

Data quality describes how fit data is for its intended use. Commonly used dimensions include accuracy, completeness, consistency, timeliness, validity, and uniqueness. Each dimension can be measured with rules, such as the percentage of customer records that contain a valid postal code.

Question 18

Why is uniqueness a central quality dimension in MDM?

Accepted Answer

Uniqueness means each real-world entity has exactly one record in the master data, so duplicates do not distort counts, billing, or analysis. In MDM, uniqueness is the outcome that matching and consolidation aim to achieve. Duplicate customers can receive two invoices, while duplicate products can distort inventory totals.

Question 19

What is a data quality rule, and how is it applied?

Accepted Answer

A data quality rule is a testable condition that a value or record must satisfy, such as a required field not being blank or a date falling within a plausible range. Rules are applied when data is entered, when it is loaded from source systems, and during periodic profiling. Failures go to a steward.

Question 20

What is data profiling, and why do MDM teams perform it early?

Accepted Answer

Data profiling is the analysis of existing data to discover its structure, content, and quality, including value distributions, null rates, and patterns. MDM teams profile early because it reveals duplicates, inconsistent formats, and hidden business rules before design decisions are made. Results shape matching logic, data models, and the size of the cleanup effort.

Question 21

How does standardization differ from cleansing in data preparation?

Accepted Answer

Standardization differs from cleansing because it changes valid values into one agreed format, while cleansing corrects or removes values that are wrong. Cleansing fixes a misspelled city or deletes a corrupted record, while standardization writes every state as its two-letter code. Both steps improve matching accuracy.

Question 22

What is a data quality metric, and how should teams define one?

Accepted Answer

A data quality metric is a measurable indicator of a quality dimension, expressed as a count or percentage over a defined population. A good metric states what is measured, which records are in scope, the calculation formula, and the target threshold. Vague goals are hard to track.

Question 23

How can a team measure duplicate rates in customer data?

Accepted Answer

A team measures duplicate rate by running match rules across the customer population and dividing the number of records identified as duplicates by the total number of records. Sampling a few hundred matched pairs and having stewards confirm them also estimates precision. The result shows how much consolidation work remains.

Question 24

Why should data quality targets be tied to business outcomes?

Accepted Answer

Quality targets tied to business outcomes focus effort on the errors that cost money or trust, rather than on perfection for its own sake. For example, a retailer might prioritize accurate shipping addresses for customers, because wrong addresses cause failed deliveries and refunds. Linking targets to outcomes also helps secure funding and steward attention.

Question 25

What is data integration in the context of MDM?

Accepted Answer

Data integration is the process of moving, transforming, and combining data from source systems so that it can be used together. In MDM, integration connects operational applications to the hub, feeding source records in and distributing mastered values back out. Common integration patterns include batch loads, real-time messaging, and application programming interface calls.

Question 26

What is the difference between batch and real-time synchronization of master data?

Accepted Answer

Batch synchronization moves changes between systems on a schedule, such as nightly, so records can be out of date between runs. Real-time synchronization pushes each change as it happens, which keeps systems closer in step but demands more reliable messaging and error handling. The right choice depends on how quickly users need updates.

Question 27

What is a change event in a master data hub?

Accepted Answer

A change event is a message that records a create, update, merge, or deactivate action on a master record, along with its identifier and timestamp. Publishing change events lets subscribing systems react to updates without polling the hub. Events create an audit trail, which helps explain how a golden record reached its current state.

Question 28

What is an identifier crosswalk, and why do MDM systems need one?

Accepted Answer

An identifier crosswalk is a mapping table that links each source system's local key to the master identifier assigned to the same entity. It is needed because applications keep their own account numbers and codes, and without the mapping, a source record cannot be traced back to its golden record or updated correctly.

Question 29

What happens when two source systems update the same master attribute at the same time?

Accepted Answer

A conflict occurs, and MDM must resolve it through predefined rules rather than by whichever update arrives last. Common approaches include ranking systems by trust for each attribute, using the most recent timestamp with a verified source, or routing the case to a steward. Consistent handling prevents silent overwrites of valid data.

Question 30

What is a data dictionary's role in master data management?

Accepted Answer

A data dictionary documents each master data attribute, including its business definition, format, allowed values, owner, and the systems that source it. It gives integration teams and stewards a shared vocabulary so that a field called status means the same thing everywhere. Without it, teams often build conflicting mappings from assumptions.

Question 31

Why is an audit trail important for master data changes?

Accepted Answer

An audit trail records who changed a master record, what changed, when it changed, and why, if the reason was captured. Auditors, stewards, and analysts use it to investigate errors, prove compliance with internal policy, and roll back incorrect merges. Without it, a bad change can spread unnoticed through many systems.

Question 32

How should an MDM team handle an incorrect merge after it has reached downstream systems?

Accepted Answer

The team should stop further propagation, then use the audit trail and source identifiers to split the records back into their original entities. Downstream systems are then notified of the corrected identifiers so they can reverse any affected transactions. Then the team should review the match rule that caused the error and adjust it.

Question 33

What are the main steps in an MDM program roadmap?

Accepted Answer

A typical roadmap starts with defining business goals and the domains in scope, then assessing current data quality and sources. Next come designing the data model and governance, selecting a style and platform, and loading and matching data. The program closes with rollout to consuming systems, then ongoing stewardship.

Question 34

Why should an MDM program start with a limited domain, such as customers?

Accepted Answer

Starting with one well-chosen domain limits scope, builds early proof of value, and lets the team refine governance and matching before expanding. Customers are a common first choice because the business impact of duplicates is visible and many systems depend on the same data.

Question 35

How do you build a business case for an MDM program?

Accepted Answer

A business case links MDM to measurable outcomes, such as fewer duplicate mailings, faster customer onboarding, or more accurate financial reporting. It estimates costs for tools, integration, and stewardship, then compares them with the expected savings and risk reduction. Presenting a small pilot's results alongside the case makes the claims more credible.

Question 36

How should stewardship workloads be organized as the MDM program scales?

Accepted Answer

Stewardship work should be assigned by domain or business unit, with clear escalation paths for disputed records. Stewards need queues that rank exceptions by business impact, so the most costly errors are handled first. As volume grows, teams add automated rules for clear-cut cases and reserve human review for ambiguous ones.

Question 37

What key performance indicators show whether an MDM program is succeeding?

Accepted Answer

Useful indicators include the duplicate rate over time, the percentage of records meeting quality thresholds, the share of systems consuming the golden record, and the time needed to resolve stewardship exceptions. Tracking trends matters more than single snapshots. Targets should be agreed with business owners before launch, so results can be judged fairly.

Question 38

How does an MDM program handle changes to its data model over time?

Accepted Answer

Changes to the data model should go through a formal change process that assesses the impact on match rules, integrations, reports, and stewards. Versioning the model and documenting each change lets consuming systems adjust in a controlled way. Skipping this process often breaks downstream feeds without warning.

Question 39

What is a common mistake when launching an MDM program?

Accepted Answer

A common mistake is treating MDM as a one-time cleanup project rather than an ongoing discipline. Once the initial load is finished, nobody owns new data entry, so duplicates and inconsistencies return. Sustained success depends on governance, stewardship, and entry controls that keep the golden record accurate over time.

Question 40

How can an MDM program support privacy requirements for personal data?

Accepted Answer

An MDM program supports privacy by recording where personal data lives, so an organization can locate every copy of a customer's information when a request arrives. It can also store consent status and preferences centrally, so downstream systems respect them. Access rules applied to the hub limit who can view sensitive attributes.

Master Data Management Flashcards

1Core Concepts of Master Data

2Architecture, Matching, and Survivorship

3Data Quality and Measurement

4Integration and Synchronization

5Program Rollout and Operations

Learning Resources