Materials Master Data and Information Governance

Introduction and Context

Materials management increasingly depends on the quality of information rather than on physical storage alone. An enterprise resource planning (ERP) system may contain millions of records describing raw materials, components, packaging, finished goods, services, suppliers, customers, locations, prices, units of measure and relationships among these objects. These records are not merely administrative descriptions. They drive purchasing, inventory valuation, production planning, warehouse operations, quality control, maintenance, sales, regulatory reporting and analytics.

A material master record is the controlled, enterprise-wide representation of an item that an organisation buys, stores, manufactures, sells, consumes or otherwise manages. If the same physical item is represented by several inconsistent records, the organisation may experience duplicated inventory, fragmented demand, incorrect replenishment, purchasing leakage, inaccurate costing and unreliable management reports. Conversely, if distinct items are incorrectly merged into one record, safety, quality, legal and operational risks may result.

Materials master data governance therefore concerns both data design and organisational accountability. It specifies what a material is, how it is identified, who may create or change it, how its quality is measured, how long it is retained and how it is exchanged with other systems. Governance must connect business policy, information architecture, ERP configuration, supplier collaboration and regulatory obligations.

Learning Objectives

After studying this chapter, the learner should be able to:

1. Explain the role and structure of materials, supplier, bill-of-materials, classification and units-of-measure data.
2. Distinguish master data from transactional, reference and metadata.
3. Design coding and naming policies that support uniqueness, searchability and interoperability.
4. Establish ownership, stewardship, approval, quality and lifecycle controls.
5. Analyse the effect of poor master data on planning, procurement, inventory and financial performance.
6. Evaluate ERP and digital technologies used for data governance.
7. Recommend a governance approach for a complex materials-management environment.

1. The Information Architecture of Materials Management

1.1 Master, transactional, reference and metadata

Master data describes relatively stable business objects and their essential attributes. Examples include a material, supplier, plant, storage location or customer. Master data is reused across many transactions.

Transactional data records events. A purchase order, goods receipt, stock transfer, production order and invoice are transactions that refer to master-data objects.

Reference data provides controlled values used in processes, such as currency codes, country codes, material types, incoterms, tax categories and units of measure.

Metadata describes data itself. A field definition, data owner, permissible values, source system, retention rule and last-change timestamp are metadata.

The distinction is important because each category requires different controls. A purchase order cannot be made reliable if its material or supplier master is unreliable; however, correcting a transaction is not the same as correcting the underlying master record.

1.2 The material master as a multidimensional record

A material master normally contains a global core and organisational extensions. The core may include material number, description, base unit, material type, gross weight, net weight, dimensions and classification. Organisational views may include purchasing group, valuation class, standard or moving-average price, planning parameters, warehouse data, quality-inspection settings and sales data.

The same material can therefore have one identity but several context-dependent attributes. A material may be purchased by one plant, manufactured by another and stocked in multiple locations. Governance must distinguish attributes that are globally common from those that legitimately vary by plant, company code, warehouse or sales organisation.

A useful conceptual structure is:

2. Material Identification, Coding and Naming

2.1 Identification principles

An identification scheme should provide one and only one enterprise identity for a materially distinct item. “Materially distinct” means that a difference affects procurement, use, form, fit, function, quality, safety, legal treatment, valuation or planning. A different colour may require a separate material in one industry but only a characteristic value in another. The decision must be based on business consequences rather than on visual difference alone.

A strong policy normally includes:

• Uniqueness: each approved material has one identifier.
• Persistence: the identifier is not recycled after deletion or discontinuation.
• Neutrality: the identifier does not embed attributes likely to change, such as supplier, price or warehouse.
• Non-significance: where feasible, use a meaningless sequential or system-generated number rather than a complex intelligent code.
• Searchability: descriptions, aliases and classification values support human retrieval.
• Cross-reference: supplier, manufacturer and legacy identifiers are retained as controlled references.

An intelligent code such as “RM-STEEL-304-2MM” appears informative, but it becomes problematic when the specification changes, when abbreviations are interpreted differently or when the code length is insufficient. A neutral identifier, combined with governed attributes, is generally more durable. Intelligent codes may still be justified in tightly controlled environments, provided that their rules are documented and change-resistant.

2.2 Description standards

A material description should be concise, consistent and semantically rich. A common pattern is:

noun + modifying characteristics + size or grade + form

For example: “Bolt, hexagonal head, M12 × 60 mm, stainless steel A2.” The order, abbreviations, punctuation, units and spelling should be standardised. Descriptions should not contain supplier names, temporary project references or ungoverned marketing language unless those are genuine identifying characteristics.

A description standard should define controlled abbreviations, character limits, language rules, pluralisation, decimal conventions and prohibited terms. Search aliases and manufacturer part numbers should be stored in separate fields rather than hidden in free text.

2.3 Duplicate prevention and rationalisation

Duplicate detection combines deterministic and similarity-based methods. Exact matching can compare manufacturer part number, supplier part number and key attributes. Similarity methods can compare tokenised descriptions, synonyms, unit normalisation and characteristic values. Automated similarity scores should support, not replace, expert review because two descriptions may be linguistically similar but technically different.

A rationalisation decision usually has three outcomes:

1. Merge: records represent the same item and one surviving identity is selected.
2. Retain separately: differences have operational, legal, quality or financial significance.
3. Block or retire: the record is not to be used for new transactions, while historical traceability is preserved.

Deletion is rarely appropriate because historical transactions, audit trails and statutory records may depend on the identity.

3. Supplier Master Data and Business-Partner Governance

Supplier data should be designed around the legal entity and its business relationships. A supplier may have several addresses, payment locations, manufacturing sites, bank accounts, tax registrations and purchasing relationships. Governance must prevent the creation of separate supplier records merely because an address, contact or remittance instruction differs.

Important supplier attributes include legal name, registration number, tax identifiers, registered and operational addresses, bank information, payment terms, currency, incoterms, contact roles, sanctions status, diversity status, quality approvals and risk classification. Sensitive fields, especially banking and personal contact data, require access control and privacy protection.

A supplier master should distinguish:

• the legal entity that carries contractual responsibility;
• the supplier site from which goods are shipped;
• the ordering address and remittance address;
• the manufacturer, which may differ from the distributor; and
• the supplier-material relationship, including supplier part number, minimum order quantity, lead time and validity.

Supplier onboarding should include identity verification, tax validation, sanctions screening where required, bank-account verification, conflict-of-interest declarations and risk-based approval. Changes to bank details should require independent verification because fraudulent payment-redirection attacks frequently exploit weak master-data controls.

4. Bill of Materials as Governed Product Knowledge

A bill of materials (BOM) represents the components, quantities and relationships required to make, maintain or sell a parent item. It is not simply a list. It is a controlled product definition with effectivity, revision, alternative structures, scrap assumptions, phantom assemblies and sometimes operation references.

A single-level BOM identifies immediate components. A multilevel BOM expands subassemblies recursively until purchased or raw-material items are reached. A manufacturing BOM may differ from an engineering BOM because manufacturing requires shop-floor structure, operation sequence or consumable information. A service or maintenance BOM may represent an asset’s installed structure rather than production requirements.

Key governance fields include:

• parent material and component material;
• quantity per assembly and component unit;
• revision or change number;
• effective-from and, where necessary, effective-to dates;
• engineering change order or approval reference;
• alternative BOM identifier;
• phantom, bulk or non-stock indicator;
• scrap or yield assumptions; and
• plant or production-version applicability.

For a simple deterministic assembly, the gross component requirement for component i is:

where Gᵢ is the gross requirement for component i, Qₚ is the planned quantity of parent assemblies, and qᵢ is the BOM quantity of component i per parent. This relationship assumes a fixed BOM, no scrap, no substitutions and no yield loss. If a component has a scrap fraction sᵢ, the requirement may be adjusted as:

where sᵢ is expressed as a decimal between 0 and 1. The equation is applicable only when scrap is represented as a proportional loss and is not appropriate for fixed setup scrap or complex yield distributions.

Worked example. A production plan requires 500 assemblies. Each assembly uses 4 brackets, and expected bracket scrap is 2%. The gross requirement is:

G_i = (500 × 4) ÷ (1 − 0.02) = 2,040.82

Because brackets are purchased in whole units, the operational requirement must be rounded up to 2,041 units, subject to purchasing-pack constraints. The calculation illustrates why BOM quantity, scrap policy and rounding rules must be governed together.

BOM changes should follow configuration-management discipline. A change should identify the reason, approving authority, affected plants and orders, effective date, replacement component and disposition of existing stock. Without effectivity control, the same product may be planned, purchased and manufactured against incompatible definitions.

5. Classification and Attribute Management

Classification supplements the material number and description with structured characteristics. A class for electrical cable may include conductor material, cross-sectional area, insulation type, voltage rating, temperature rating and colour. A class for chemicals may include concentration, hazard category, flash point and regulatory identifiers.

Classification enables faceted search, technical comparison, spend analysis, substitution and compliance reporting. It also supports interoperability when characteristics are mapped to industry standards or customer schemas.

A governed classification model defines:

• class types and their intended scope;
• characteristic names and definitions;
• data types, permissible values and units;
• whether a characteristic is mandatory;
• synonyms and translations;
• validation rules and evidence requirements; and
• ownership and change procedures.

Free-text characteristics undermine comparison. For example, “stainless,” “SS,” and “inox” may refer to the same concept, while “304” and “316” are materially different grades. Controlled values and semantic definitions are therefore more valuable than merely adding more fields.

Classification should avoid unnecessary granularity. Excessive attributes increase maintenance cost, slow item creation and encourage users to enter placeholders. The appropriate design balances search value, regulatory need, engineering significance and lifecycle cost.

6. Units of Measure and Quantity Integrity

Units of measure are a frequent source of silent ERP error. A material may be purchased by the box, stored by the piece, consumed by the metre and priced by the kilogram. The system must record a base unit and controlled conversion relationships.

For a linear conversion, if one purchasing unit contains n base units, then:

where Qₐ is the quantity in the alternate unit, Qᵦ is the equivalent quantity in the base unit, and n has units of base units per alternate unit. The relationship is valid only where the conversion is stable and unambiguous.

Example. One carton contains 24 pieces. A receipt of 15 cartons equals:

Q_b = 15 × 24 = 360 pieces.

Weight, volume and length conversions require additional care. A conversion between volume and mass requires density, and density may vary with temperature or composition. A conversion between rolls and metres may depend on roll length and cannot be treated as universally fixed if roll sizes vary. In such cases, the unit relationship must be item-specific, batch-specific or measured at transaction time.

Governance should specify decimal precision, rounding direction, pack-size constraints, ISO or industry codes, language labels and whether a conversion is fixed, approximate or prohibited. Unit errors can distort stock levels, MRP, freight calculations, safety stock, invoices and regulatory declarations.

7. Data Quality: Dimensions, Metrics and Controls

Data quality is fitness for intended use. Common dimensions include:

A quality score should not be reduced to one number without context. A record can be complete but inaccurate, or accurate but obsolete. Metrics should be defined by process and risk. Useful measures include duplicate rate, mandatory-field compliance, percentage of records with expired certificates, average approval time, change rejection rate and incidence of transaction failure caused by master data.

A simple completeness rate for a field across N applicable records can be expressed as:

where N꜀ is the number of records containing an acceptable value and N is the number of applicable records. This measure assumes that “acceptable” has been defined; a non-blank placeholder should not count as complete.

Preventive controls are preferable to downstream cleansing. Examples include duplicate checks at creation, controlled vocabularies, field-level validation, approval workflows, segregation of duties and integration testing. Detective controls include exception reports, reconciliation, sampling, supplier review and analytics.

8. Governance Operating Model

Governance assigns decision rights. A practical model separates the following roles:

• Data owner: accountable for policy, risk and business meaning.
• Data steward: manages definitions, quality, workflow and issue resolution.
• Data custodian: operates technical platforms, security, backup and integration.
• Process owner: ensures that data supports an end-to-end process.
• Data consumer: uses data and reports defects.
• Governance council: resolves cross-functional conflicts and approves standards.

Ownership should be assigned by data domain and attribute, not merely by system. Procurement may own supplier commercial attributes, engineering may own technical specifications, finance may own valuation rules, and quality or regulatory functions may own compliance characteristics. Shared ownership without a final decision-maker leads to delay and inconsistent standards.

A controlled material-creation workflow commonly includes request, duplicate search, attribute completion, technical review, procurement or finance review, approval, system creation, replication and post-creation monitoring. Emergency creation should be exceptional, time-limited and retrospectively reviewed.

Change management should record old and new values, reason, requester, approver, effective date and affected interfaces. Access should follow least privilege. A person who requests a supplier bank change should not be the sole person who approves and releases payment against that change.

9. Lifecycle Management and Retention

Master records pass through stages such as proposed, active, restricted, obsolete, blocked and archived. Lifecycle states must have operational meaning. “Blocked for purchasing” may still permit consumption of existing stock; “blocked for all transactions” has a different effect.

Lifecycle rules should address:

• introduction and approval;
• extensions to new plants or warehouses;
• temporary materials and project items;
• engineering revisions;
• supersession and substitution;
• discontinuation and remaining stock;
• legal and audit retention; and
• archival or destruction.

A material should not be retired solely because it has had no recent movement. Critical spare parts may have low usage but high consequence of unavailability. Obsolescence decisions should consider demand, equipment installed base, contractual obligations, hazardous properties, replacement compatibility and retention requirements.

10. Interoperability and ERP Integration

Materials data moves among ERP, product lifecycle management, warehouse management, manufacturing execution, laboratory, supplier portals, e-commerce, transportation and analytics platforms. Interoperability requires shared semantics, identifiers, code lists, interfaces and error-handling rules.

Important design principles include:

1. Define the system of record for each attribute.
2. Use canonical identifiers and maintain cross-reference tables.
3. Publish data contracts specifying fields, formats, units, cardinality and timing.
4. Validate data at the source and at integration boundaries.
5. Make interfaces idempotent where possible, so repeated messages do not create duplicate objects.
6. Record rejected messages and provide accountable resolution queues.
7. Distinguish a missing value from a value that is not applicable.

An ERP migration should not simply copy every legacy record. Legacy systems often contain duplicates, obsolete descriptions, incompatible units and undocumented local codes. A migration programme should profile data, define the target model, map and cleanse records, test conversions, reconcile counts and values, obtain business sign-off and retain a historical cross-reference.

11. Technology, Automation and AI

Modern governance platforms may provide workflow, validation, duplicate detection, reference-data management, data catalogues, lineage and application programming interfaces. Machine-learning methods can identify likely duplicates by comparing description tokens, supplier references and technical attributes. Natural-language processing can propose standardised descriptions or classify free text.

These tools should be treated as decision support. Automated recommendations can reproduce historical bias, misinterpret technical terminology or merge items that are merely similar. High-risk decisions require human review, explainable evidence and an audit trail. Generative AI may assist with field suggestions, documentation and exception summarisation, but confidential supplier, pricing and personal data must be protected. AI outputs should never bypass approval, segregation of duties or regulatory controls.

12. Illustrative Case: Industrial Equipment Manufacturer

Background

A fictional industrial-equipment manufacturer operates three plants and an ERP system introduced through acquisitions. It reports 180,000 material records, although procurement estimates that fewer than 90,000 distinct items are actively required. Each plant uses different abbreviations and units. Engineering maintains BOMs in a product-lifecycle system, while procurement maintains supplier part numbers in spreadsheets.

Problem

The organisation experiences duplicate purchases, frequent MRP exceptions and inconsistent inventory valuation. A bearing is purchased by one plant in pieces and by another in boxes. Several apparently identical bearings differ in seal type and temperature rating. A BOM revision is released to engineering but not replicated to production planning.

Analysis

A data-profiling exercise identifies 21% probable material duplicates, 8% missing base units, 14% uncontrolled descriptions and 6,400 supplier records lacking validated tax identifiers. Interviews reveal that no group owns the definition of a material, and local users create records to meet urgent production needs.

Approach and solution

The company establishes a material-data council chaired by operations, with engineering, procurement, quality, finance, IT and plant representation. It adopts neutral identifiers, a controlled description grammar and a classification model for bearings, fasteners and electrical items. Engineering owns technical characteristics; procurement owns supplier relationships; finance owns valuation attributes; and IT operates the workflow and interfaces.

A duplicate-screening service proposes candidates, but a technical steward decides whether records are merged. Unit conversions are validated against packaging specifications. BOMs receive revision and effective-date controls, and an interface reconciliation report compares approved engineering structures with production structures daily. New records require a duplicate search, mandatory characteristics and approval; emergency records expire from active use unless regularised within ten working days.

Outcome

After six months, the company retires or consolidates 32,000 records, validates critical supplier identities and standardises 1,100 unit relationships. MRP exceptions decline, purchasing users can compare supplier offers more reliably, and inventory reports become more credible. The improvement is not attributed to cleansing alone: decision rights, process controls and integration monitoring prevent the same problems from returning.

Learning points

The case demonstrates that master-data quality is an operating capability. Technology identifies patterns, but domain experts decide meaning. Governance must cover definitions, workflow, lifecycle, integration and accountability. Local flexibility should be preserved only where a genuine business requirement exists.

13. Challenges, Ethics and Regulatory Considerations

Master-data decisions can affect people, suppliers and communities. Supplier classification may influence access to contracts; risk scores may affect payment or onboarding; and personal contact information is subject to privacy obligations. Organisations should collect only necessary personal data, define retention periods, restrict access and provide correction mechanisms.

Regulatory requirements may apply to tax data, hazardous substances, product traceability, conflict minerals, export controls, sanctions, medical devices, food ingredients or environmental reporting. A governance programme should identify the applicable jurisdiction, authoritative evidence, accountable owner and audit-retention period. It should not assume that a field is legally sufficient merely because it is populated.

Common limitations include the cost of stewardship, conflict between global standards and local needs, changing product specifications, supplier data unreliability and the difficulty of quantifying avoided errors. Governance can also become bureaucratic if every low-risk change follows the same approval path. Risk-based controls are therefore preferable: high-risk chemicals, banking attributes and regulated products deserve stronger evidence and segregation than low-risk descriptive aliases.

Best-Practice Framework

A mature materials-information governance programme should:

1. Define business terms and data ownership before configuring fields.
2. Design a neutral, persistent identification policy.
3. Separate identity, description, classification and supplier references.
4. Establish authoritative units, conversions and rounding rules.
5. Govern BOM revisions, effectivity and substitutions.
6. Use preventive validation and duplicate detection at creation.
7. Apply risk-based approval, access and segregation of duties.
8. Measure quality by process-relevant dimensions.
9. Maintain lineage, audit trails and integration reconciliation.
10. Treat migration and cleansing as continuing capabilities rather than one-time projects.
11. Use AI and automation with human accountability and explainability.
12. Review governance performance through operational outcomes, not record counts alone.

Chapter Summary

Materials master data provides the stable information foundation for procurement, inventory, planning, production, quality, finance and analytics. Effective governance distinguishes master, transactional, reference and metadata; defines material identity; standardises descriptions; controls supplier and supplier-material relationships; and treats BOMs as versioned product knowledge. Classification makes technical attributes searchable and comparable, while governed units of measure protect quantity integrity.

Data quality must be assessed through accuracy, completeness, consistency, uniqueness, timeliness, validity and traceability. Ownership, stewardship, workflow, segregation of duties and lifecycle states convert policy into operating control. Interoperability requires shared semantics, systems of record, canonical identifiers, data contracts and reconciliation. ERP implementation, migration and AI-assisted cleansing are valuable only when supported by accountable business decisions. The central managerial lesson is that master-data quality is not an IT-cleaning exercise; it is a cross-functional governance capability linked directly to operational and strategic performance.

Key Terms

Master data: Stable, reusable data describing business objects such as materials and suppliers.

Transactional data: Data recording business events such as purchase orders and goods receipts.

Reference data: Controlled codes and values used to standardise transactions.

Metadata: Data describing the meaning, origin, structure or control of other data.

Material master: The governed ERP representation of an item managed by the organisation.

Neutral identifier: A persistent identifier that does not encode changeable attributes.

Intelligent code: An identifier whose characters convey business meaning.

Data owner: Person accountable for the policy and business meaning of a data domain.

Data steward: Person responsible for operational quality, definitions and issue resolution.

Bill of materials: A governed structure describing components and quantities required for a parent item.

Effectivity: The time, revision or condition under which a data structure is valid.

Classification: Structured grouping of objects using governed characteristics and values.

Unit of measure: A standard quantity reference such as piece, kilogram or metre.

Duplicate record: An unintended second representation of the same business object.

Data lineage: The traceable history of data origin, transformation and movement.

Data contract: An agreed specification for data exchanged between systems.

Lifecycle state: A controlled status such as proposed, active, blocked or obsolete.

Canonical identifier: A common identity used to connect representations across systems.

Review Questions

Short Answer Questions

1. Distinguish master data from transactional data and give one example of each.
2. Why are neutral material identifiers generally more durable than intelligent codes?
3. What is the purpose of effectivity control in a bill of materials?
4. Name four important dimensions of data quality.
5. Why may a volume-to-mass unit conversion require more than a fixed ratio?

Long Answer Questions

1. Explain how material identification, description standards, classification and duplicate prevention work together in an ERP environment.
2. Discuss the governance requirements for supplier master data, including legal identity, supplier sites, banking information and segregation of duties.
3. Analyse the role of BOM governance in production planning, engineering change management and inventory control.
4. Design a data-governance operating model for a multinational manufacturer with global and local material requirements.
5. Evaluate the benefits and limitations of using AI for master-data cleansing and classification.

Analytical / Application Questions

1. An organisation receives 15 cartons containing 24 pieces each. Explain the unit conversion and identify two governance controls needed to prevent quantity errors.
2. A company has two bearing records with nearly identical descriptions, but one has a higher temperature rating and a different seal type. Should the records be merged? Justify your decision using material-identity principles.
3. A production plan requires 500 assemblies, each using 4 brackets, with 2% proportional scrap. Calculate the gross requirement, explain the rounding decision and state the assumptions behind the calculation.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes master data from transactional data?

A. Master data records individual business events, while transactional data defines stable objects.

B. Master data describes reusable business objects, while transactional data records events involving those objects.

C. Master data is always numeric, while transactional data is always textual.

D. Master data is stored outside ERP systems, while transactional data is stored inside them.

Correct Answer: B. Master data describes reusable business objects, while transactional data records events involving those objects.

Question 2

Which is the strongest reason for using a neutral material identifier?

A. It guarantees that no description is required.

B. It embeds supplier and price information automatically.

C. It remains useful when descriptive attributes or suppliers change.

D. It eliminates the need for duplicate detection.

Correct Answer: C. It remains useful when descriptive attributes or suppliers change.

Question 3

A BOM requires 500 assemblies, each using 4 brackets, with 2% proportional scrap. What gross quantity is required before whole-unit rounding?

A. 1,960.00

B. 2,000.00

C. 2,040.82

D. 2,080.00

Correct Answer: C. 2,040.82

Question 4

Which role is primarily accountable for the business meaning and policy of a data domain?

A. Data owner

B. Data consumer

C. System operator

D. Interface scheduler

Correct Answer: A. Data owner

Question 5

What is the main purpose of classification in a material master?

A. To replace all material identifiers

B. To provide structured characteristics for search, comparison and analysis

C. To record purchase-order approval history

D. To calculate supplier bank balances

Correct Answer: B. To provide structured characteristics for search, comparison and analysis

Question 6

One carton contains 24 pieces. How many pieces are represented by 15 cartons?

A. 39 pieces

B. 240 pieces

C. 360 pieces

D. 480 pieces

Correct Answer: C. 360 pieces

Question 7

Which control is most appropriate for a change to supplier bank details?

A. Allow unrestricted self-approval by the requester

B. Require independent verification and segregation of duties

C. Delete the previous supplier record immediately

D. Permit changes only through free-text notes

Correct Answer: B. Require independent verification and segregation of duties

Question 8

Which statement about volume-to-mass conversion is correct?

A. It always uses a universal fixed ratio.

B. It may require density and therefore may vary by material or condition.

C. It is impossible in an ERP system.

D. It never affects inventory valuation.

Correct Answer: B. It may require density and therefore may vary by material or condition.

Question 9

What does BOM effectivity control primarily establish?

A. The user’s password age

B. The conditions or period in which a BOM revision is valid

C. The supplier’s payment currency only

D. The physical location of every warehouse

Correct Answer: B. The conditions or period in which a BOM revision is valid

Question 10

Why should AI recommendations for duplicate materials normally receive human review?

A. AI cannot compare text or attributes.

B. Similarity does not prove technical identity, and incorrect merging may create operational risk.

C. Human review prevents all future data changes.

D. AI tools can only process financial records.

Correct Answer: B. Similarity does not prove technical identity, and incorrect merging may create operational risk.