How do you ensure every department in a company uses the same definition of 'customer' — or the same list of countries or product categories? Master data management (MDM) and reference data solve this problem by creating a single, trusted source of information that everyone agrees to use. For the DA0-002 exam, you need to understand the difference between master data and reference data, why MDM matters for data quality, and how it prevents the costly chaos of inconsistent datasets across an organisation.
Jump to a section
A simple way to picture Master Data Management and Reference Data
A city's public library system holds thousands of books, each with a unique barcode and a catalogue record. The catalogue record is the master data for that book — it contains the title, author, ISBN, publication year, and subject headings. This record never changes, no matter how many times the book is borrowed, returned, or moved to a different shelf. The library's reference data is the list of valid subject headings (e.g., 'Science Fiction', 'Biography', 'History') and the list of authorised publisher codes. These reference lists are kept in a separate, controlled file that every librarian uses to ensure consistency when categorising new books.
Now, imagine a librarian who decides to create her own subject heading 'Sci-Fi' while another uses 'Science Fiction'. Over time, the catalogue becomes a mess — you can't reliably find all the sci-fi books. Master data management is the set of rules, processes, and technology that ensures every librarian uses the same master record for each book and the same reference data for categorisation. It replaces chaos with a single source of truth. The library's master data is the 'golden record' for each item; the reference data is the approved vocabulary for describing it. Without both, the catalogue is unreliable, and patrons lose trust in the system.
Master data management (MDM) is the practice of creating and maintaining a single, authoritative source of core business entities — things like customers, products, employees, and suppliers. These entities, called 'master data', are the nouns of your business: the things you sell, the people you serve, and the resources you use. Master data is relatively stable over time (a customer's name changes rarely), but it's absolutely essential for daily operations. Without MDM, the same customer might be recorded as 'John Smith' in the sales system, 'J. Smith' in the billing system, and 'Jonathan Smith' in the marketing database — leading to missed invoices, duplicate mailings, and terrible reporting.
Reference data, on the other hand, is a type of master data that provides standardised, predefined values for categories or codes. Think of it as a controlled vocabulary or a lookup list that limits what can be entered into a field. Common examples include country codes (US, UK, DE), currency codes (USD, EUR, GBP), product categories ('Electronics', 'Clothing'), status values ('Active', 'Inactive'), and industry standard codes such as UNSPSC (United Nations Standard Products and Services Code). Reference data changes very infrequently and is usually maintained by an external standards body or an internal governance committee.
Why does this matter for data quality? Inconsistent master and reference data leads to 'dirty data' — records that are incomplete, duplicated, or contradictory. For example, if your sales team uses 'USA' and your shipping team uses 'United States', a report on sales by country will show two separate entries for the same country. This makes analytics unreliable and can cause operational mistakes (like shipping to the wrong address). MDM solves this by creating a 'golden record' — a single, clean, consolidated view of each master data entity. This golden record is the version of the truth that every system and department agrees to use.
So how does MDM work in practice? It typically involves the following stages:
Data identification: finding all the different sources of master data across the organisation (e.g., CRM, ERP, billing system)
Data matching: using algorithms to identify which records in different systems refer to the same real-world entity (e.g., matching 'John Smith' and 'J. Smith' by address or phone number)
Data consolidation: merging matched records into a single golden record, resolving conflicts (e.g., which address is most current?)
Data governance: establishing rules and roles for who can create, update, or delete master data, and who approves changes
Data distribution: sharing the golden record back to all connected systems to keep them synchronised
Reference data management has a slightly different process. Since reference data is typically a fixed list of values, the focus is on version control and access control. You define the list of allowed values (e.g., official country codes from ISO 3166), store it in a central repository, and ensure all systems pull from that repository rather than maintaining their own copies. When a new value is needed (e.g., a new currency code), a governance board approves the change, and the updated list is distributed to all systems.
Before MDM, companies often managed master data in spreadsheets or in separate copies within each application — a practice called 'siloed data management'. This led to endless reconciliation work, duplicate records, and poor data quality. MDM replaces this chaos with a systematic, automated approach that saves time, reduces errors, and enables reliable analytics. For the DA0-002 exam, focus on knowing the definition of master data vs. reference data, the benefits of MDM, and the types of data that qualify as master data (customers, products, employees, locations). You should also recognise that reference data is a subset of master data — it's the standardised codes and categories used to describe the master entities.
Identify master data entities
Determine which core business entities (e.g., customer, product, employee, supplier) exist across multiple systems. This is the foundation of any MDM project. Without this step, you don't know what to manage.
Audit and profile source systems
Examine each system (CRM, ERP, billing, etc.) to understand how master data is currently stored, what fields exist, and what quality issues are present (e.g., missing values, duplicates). This reveals the scope of the problem.
Match and link records
Use matching algorithms to identify which records in different systems refer to the same real-world entity. This often involves comparing fields like name, address, email, or phone number. Accurate matching is key to creating a reliable golden record.
Consolidate into a golden record
Merge all matched records into a single, clean, authoritative version — the golden record. Conflicts (e.g., different addresses) are resolved based on predefined rules (e.g., most recent update wins). This golden record becomes the single source of truth.
Govern and distribute master data
Establish data governance rules (who can edit, how changes are approved) and then distribute the golden record back to all connected systems so they stay synchronised. Without distribution, the master data remains isolated and ineffective.
Imagine you work as a junior data analyst for 'GreenLeaf Furniture', a mid-sized company that sells office furniture online and through a physical showroom. GreenLeaf has three separate systems: an e-commerce website, a warehouse inventory system, and a customer support ticketing platform. Each system has its own customer database, its own product catalogue, and its own set of status codes. This is a recipe for data disasters — and you've just been asked to produce a sales report by country.
Here's what you do step by step:
Identify the master data entities: you recognise that 'customer' and 'product' are the core master data entities that need to be unified. You also notice that 'country' is reference data — a standardised list of country names and codes.
Audit existing data sources: you export customer records from all three systems. The e-commerce platform lists a customer as 'Jane Doe, USA', the warehouse system has 'Jane D., US', and the support platform has 'Jane Doe, United States'. Similarly, the same product 'ErgoChair Pro' is listed as 'ErgoChair Pro' (e-commerce), 'Ergo Chair Pro' (warehouse), and 'ERG-100' (support).
Perform data matching: you use a tool that compares fields like email address, phone number, and postal address. It identifies that all three 'Jane Doe' records belong to the same person based on matching email and phone number. For the product, you match on the manufacturer's SKU (stock-keeping unit).
Create a golden record: you define the master customer record as 'Jane Doe', with address '123 Main St, USA'. You standardise the country to the ISO code 'US'. The master product record becomes 'ErgoChair Pro' with SKU 'ERG-100'.
Standardise reference data: you create a central list of allowed country codes using ISO 3166-1 alpha-2 (e.g., US, GB, DE). You also create a list of product categories ('Chairs', 'Desks', 'Accessories') and order statuses ('Pending', 'Shipped', 'Delivered'). All systems must now use exactly these values.
Govern the data: you set up a simple approval process — any change to the master customer record requires the customer's consent, and any new product category must be approved by the product team lead.
Distribute the golden record: you use an integration tool to push the unified customer and product records back to all three systems, so they are now synchronised.
The result: your sales report by country now shows clean numbers — all US customers appear under 'US', not 'USA', 'US', or 'United States'. The company can now trust the data for budgeting, marketing, and logistics. As an IT professional, your role involves using MDM tools (like Profisee, Informatica MDM, or Talend) to automate these steps, writing match rules, and ensuring data governance policies are followed. For the DA0-002 exam, you won't need to know specific tool names, but you should understand the workflow and the benefits of this approach.
The DA0-002 exam tests your understanding of master data management and reference data in several specific ways. You can expect multiple-choice questions that ask you to classify a given piece of data as master data, reference data, or transactional data. This is a common trap — candidates often confuse reference data with transactional data. Remember: transaction data describes an event (a sale, a shipment), master data describes the core entities (customer, product), and reference data is a standardised list of codes or categories (country codes, product types). Another frequent question asks you to identify the primary benefit of MDM — the correct answer is always 'improved data consistency' or 'single source of truth', not 'faster transaction processing' or 'reduced storage costs'.
Key concepts to memorise:
Definition of master data: core business entities (customer, product, employee, supplier, location) that are shared across the organisation and relatively stable
Definition of reference data: a subset of master data consisting of standardised codes/categories used to label other data (e.g., ISO country codes, currency codes, status values)
Examples of reference data: country codes, state codes, product categories, job titles, marital status values, payment method codes
Benefits of MDM: single source of truth, improved data quality, consistent reporting, reduced duplicate records, better operational efficiency
Data governance: the set of rules, roles, and processes that ensure master data is created, updated, and deleted in a controlled way
Golden record: the single, clean, consolidated version of a master data entity after matching and merging
Common reference data standards: ISO 3166 (countries), ISO 4217 (currencies), UNSPSC (products), NAICS (industries)
Exam traps to watch for:
They might list a customer's purchase history (transactional data) and ask if it's master data — it's not. The customer's name, address, and ID are master data; the purchase history is transactional.
They might describe a list of valid job titles as 'master data' — this is actually reference data. All reference data is master data, but the exam likes to test if you can distinguish between 'generic master data' (like a customer record) and 'reference data' (like a standardised list of titles).
They might include an answer option like 'MDM reduces the size of databases' — this is false. MDM may eliminate duplicates, but its primary purpose is data quality, not storage reduction.
They might test the term 'golden record' — know that it's the result of data matching and consolidation, not a copy of the original data.
Question patterns to expect:
'Which of the following is an example of reference data?' (correct answer: a list of ISO country codes)
'What is the main goal of master data management?' (correct answer: to create a single, consistent view of core business entities)
'Which type of data does a golden record represent?' (correct answer: the consolidated, authoritative version of a master data entity)
'Which of the following is NOT a benefit of MDM?' (wrong options often include 'eliminates all data errors' or 'replaces all other databases')
Spend extra time memorising the difference between master data and reference data. The exam loves this distinction.
Master data describes the core business entities — customers, products, employees, suppliers, and locations — that are shared across an organisation.
Reference data is a subset of master data that consists of standardised codes or categories, such as country codes, currency codes, and product types.
The primary goal of master data management (MDM) is to create a single, trusted 'golden record' for each master data entity, eliminating duplicates and inconsistencies.
MDM replaces siloed, inconsistent data management with a governed, automated process that improves data quality and enables reliable reporting.
Data governance is a critical component of MDM, defining who can create, update, or delete master data and under what circumstances.
The DA0-002 exam frequently tests the ability to distinguish between master data, reference data, and transactional data.
Consistent reference data ensures that every system uses the same values for categories, preventing errors like showing 'USA' and 'United States' as separate entries.
MDM does not eliminate all data errors, but it significantly reduces duplication and inconsistency across systems.
These come up on the exam all the time. Here's how to tell them apart.
Master Data
Core business entities like customers, products, and employees
Changes relatively frequently (e.g., new customer added daily)
Unique instances (each customer has a unique record)
Reference Data
Standardised codes and categories like country codes or product types
Changes very infrequently (e.g., new currency added once a year)
Limited set of predefined values (e.g., 195 country codes)
Master Data Management (MDM)
Focuses on creating and maintaining a single source of truth for master data
Involves technology processes like matching, merging, and distribution
Aims to improve data consistency and reduce duplicates
Data Governance
Focuses on rules, roles, and policies for data management
Involves human processes like approval workflows and accountability
Aims to ensure data is used and modified appropriately
Transactional Data
Records events or interactions (e.g., a sale, a shipment)
Changes frequently (e.g., every transaction adds a new row)
Depends on master data (e.g., a sale requires a customer and product)
Master Data
Describes core business entities (e.g., a customer, a product)
Changes rarely (e.g., a customer's name may change once)
Exists independently of transactions
Mistake
Master data management is just about cleaning data.
Correct
MDM is about creating and maintaining a single, authoritative source of master data through matching, merging, governance, and distribution — not just cleaning it once.
Beginners think 'management' means 'clean it up', but MDM is an ongoing process of governance and control, not a one-time cleanup project.
Mistake
Reference data is the same as transactional data.
Correct
Reference data is a static set of codes/categories used to classify data (e.g., country codes). Transactional data records events or interactions (e.g., a sale). They are fundamentally different in purpose and volatility.
Both are often stored in tables, so beginners mistakenly group them together. But transactional data changes constantly, while reference data changes rarely.
Mistake
MDM is only relevant for large corporations with complex IT systems.
Correct
MDM is valuable for any organisation that has multiple systems or data sources, even a small business using a CRM and an accounting package can benefit from consistent customer data.
People assume MDM requires expensive enterprise software, but the principles apply at any scale. The exam tests concepts, not tool cost.
Mistake
A golden record is a backup copy of the original data.
Correct
A golden record is a single, consolidated, deduplicated version of a master data entity created by matching and merging records from multiple sources. It is not a copy of one original; it's a new, authoritative version.
The word 'golden' sounds like 'golden copy' (a backup), but in MDM it means 'best version after reconciliation'.
Mistake
Master data and reference data are completely separate concepts.
Correct
Reference data is actually a type of master data. Master data includes all core business entities (customers, products, employees) and reference data is the subset that provides standardised codes or categories.
Exam questions often test this hierarchical relationship. Beginners think they are independent, but the standard categorisation puts reference data under the master data umbrella.
Reveal each answer, then mark whether you got it right. Score 60%+ to unlock the next chapter.
Master data describes the core entities of your business — customers, products, employees — that change rarely. Transactional data records events or interactions involving those entities, like a sale or a shipment, and changes frequently.
Customer address is typically master data because it is a stable attribute of the customer entity. However, if the address changes frequently (e.g., a shipping address for each order), it may be considered transactional. The exam expects you to treat the customer's primary address as master data.
A golden record is the single, consolidated, deduplicated, and most accurate version of a master data entity (like one customer) after merging records from multiple sources. It is the authoritative version that all systems should use.
While specialised MDM tools exist (e.g., Informatica MDM, Profisee), the principles can be implemented with databases and scripts. For the DA0-002 exam, you don't need to know tool names, but you should understand the workflow.
Common examples include ISO country codes (US, GB), currency codes (USD, EUR), product categories ('Electronics'), marital status values ('Single', 'Married'), and order statuses ('Pending', 'Shipped').
Without standardised reference data, different systems may use different values for the same category (e.g., 'USA' vs. 'United States'), leading to inconsistent reports and operational errors. Reference data ensures everyone uses the same labels.
You've finished Master Data Management and Reference Data. Continue through the DA0-002 study guide to build a complete picture of the exam.
Done with this chapter?