Scenario 1: Aligned Governance | ShaDey Inc. Centralized Business Glossary

1 Introduction

In this scenario, ShaDey Inc. operates with strong data governance. The Conceptual Data Model (CDM) flows cleanly into the Logical Data Model (LDM) and Physical Data Model (PDM). Business glossary terms are the single source of truth, and every data layer aligns with them. Naming, definitions, and lineage are consistent from business concept to physical column.

1. A centralized business glossary owned by the CDO with business-validated definitions.
2. Conceptual entities derive directly from glossary terms.
3. Logical attributes are specified with business rules tied to glossary definitions.
4. Physical tables and columns follow a governed naming convention and trace back to the glossary.
5. Data products for AI/ML inherit semantic consistency end-to-end.

Maturity: DAMA DMBOK Level 4–5 (Managed/Optimized) | Glossary Status: Centralized, approved, enforced | Naming Convention: Standardized across all layers | Lineage: Full column-level traceability | Data Contracts: Formal SLAs on every pipeline output

Compare with Scenario 2: Misaligned Governance to see what happens when this alignment breaks down.

2 Centralized Business Glossary

13 enterprise business terms defined, validated by domain owners, and linked to every layer of the data model. Each term has a single, authoritative definition that is consistent across CDM, LDM, and PDM.

Customer

Customer Management High

A person or organization that has purchased from or registered with ShaDey Inc. through any channel (in-store, e-commerce, wholesale).

Full Metadata

SynonymsAccount, Buyer, Patron, Shopper

Related TermsActive Customer, Loyalty Tier, Customer Type

Data OwnerVP, Customer Operations

Data StewardCustomer Data Manager

Example Usage"How many new customers acquired in Q4 via the e-commerce channel?"

Governing PolicyGDPR Art. 6 (consent), internal PII classification policy. Inactive >24 months triggers right-to-erasure review.

CDM EntityCustomer

LDM Entitycustomer: 9 attributes with validated business rules

PDM Tabledim_customer: SCD Type 2, surrogate key customer_sk

ConsistencyFully aligned across all layers

GlossaryCustomer

CDMCustomer

LDMcustomer

PDMdim_customer

Active Customer

Customer Management High

A customer who has completed at least one purchase transaction within the trailing 12 calendar months from the measurement date.

Full Metadata

SynonymsCurrent Customer, Engaged Customer

Related TermsCustomer, Lapsed Customer, Loyalty Tier

Data OwnerVP, Customer Operations

Data StewardCustomer Data Manager

Example Usage"What is our active customer count vs. same period last year?"

Governing Policy12-month window is enterprise standard. Approved by CFO and CMO. Cannot be overridden by department.

CDM EntityDerived from Customer + SalesTransaction

LDMcustomer.is_active (BOOLEAN), derived column, recalculated nightly

PDMdim_customer.is_active_flag, computed in ETL from fact_sales

ConsistencySingle definition: 12 months, enterprise-wide

GlossaryActive Customer

CDMDerived

LDMis_active

PDMis_active_flag

Product

Merchandising High

A sellable item in ShaDey Inc.'s portfolio, identified by a unique SKU, belonging to a Category within a Department.

Full Metadata

SynonymsItem, SKU, Article, Material (SAP legacy)

Related TermsCategory, Department, Brand, Variant, Supplier

Data OwnerVP, Merchandising

Data StewardProduct Master Data Manager

Example Usage"How many products have a margin below 20%?"

Governing PolicySKU format: [DEPT]-[CAT]-[SEQ]. Must have active variant and price before listing.

CDMProduct: IS-A MaterialItem, BELONGS-TO Category, HAS-MANY Variants

LDMproduct: 8 attributes with validated rules

PDMdim_product: SCD Type 2, clustered on sku

ConsistencyFully aligned. "Product" used consistently, never "Material".

GlossaryProduct

CDMProduct

LDMproduct

PDMdim_product

Sale

Sales & Revenue High

A completed purchase transaction in which a customer acquires one or more products through a channel at an agreed price.

Full Metadata

SynonymsTransaction, Purchase, Sales Transaction

Related TermsNet Sales, Order, Return, Promotion, Channel

Data OwnerHead of Sales Operations

Data StewardSales Data Analyst

Example Usage"Total sales revenue by channel for March 2026."

Governing PolicySales are distinct from Orders (intent) and Returns (reversals). Revenue recognized at point of delivery.

CDMSalesTransaction

LDMsales_transaction

PDMfact_sales: grain is one row per line item

ConsistencyAligned. Sales are always separated from Returns.

GlossarySale

CDMSalesTransaction

LDMsales_transaction

PDMfact_sales

Net Sales

Sales & Revenue High

Gross sales revenue minus returns, discounts, allowances, and sales taxes. The primary top-line revenue metric for financial reporting.

Full Metadata

SynonymsRevenue, Net Revenue

Related TermsSale, Return, Promotion, Gross Sales

Data OwnerCFO / Head of FP&A

Data StewardFinance Data Analyst

Example Usage"Net sales for the Southeast region, YTD."

Governing PolicyFormula: SUM(fact_sales.line_total) − SUM(fact_return.refund_amount) − SUM(promotion discounts). Single enterprise formula.

CDMDerived metric from SalesTransaction + Return + Promotion

LDMCalculated field with documented formula

PDMvw_net_sales (view) or agg_net_sales_daily (materialized)

ConsistencyOne formula, enterprise-wide, approved by Finance

GlossaryNet Sales

CDMDerived

LDMCalc field

PDMvw_net_sales

Inventory On Hand

Supply Chain High

The quantity of a product physically available for sale at a specific location (store or warehouse), measured in sellable units, excluding reserved, damaged, or in-transit stock.

Full Metadata

SynonymsStock on Hand, Available Inventory, SOH

Related TermsInventory, Safety Stock, Reorder Point, Days of Supply

Data OwnerVP, Supply Chain

Data StewardInventory Planning Manager

Example Usage"What is the inventory on hand for SKU ABC-123 at Store #42?"

Governing PolicyExcludes reserved, damaged, and in-transit quantities. Snapshotted daily at 23:59 UTC.

CDMInventory

LDMinventory.quantity_on_hand (INTEGER, ≥ 0)

PDMfact_inventory_snapshot.quantity_on_hand

ConsistencyAligned. "On Hand" excludes reserved/in-transit everywhere.

GlossaryInventory On Hand

CDMInventory

LDMquantity_on_hand

PDMfact_inventory_snapshot

Supplier

Procurement Medium

An external organization that manufactures, distributes, or imports products sold by ShaDey Inc., bound by a procurement contract.

Full Metadata

SynonymsVendor, Manufacturer, Distributor

Related TermsProduct, Lead Time, Quality Rating, Contract

Data OwnerVP, Procurement

Data StewardSupplier Master Data Manager

Governing PolicySupplier must pass quality audit before first PO. Rating recalculated quarterly.

CDMSupplier: PROVIDES Product, HAS-A Contract

LDMsupplier: 10 attributes

PDMdim_supplier

ConsistencyAligned. "Supplier" used consistently, never "Vendor" in models.

GlossarySupplier

CDMSupplier

LDMsupplier

PDMdim_supplier

Promotion

Merchandising Medium

A time-bound pricing incentive (discount, BOGO, coupon, markdown) applied to one or more products or categories to drive sales volume.

Full Metadata

SynonymsDeal, Campaign, Offer, Markdown

Related TermsNet Sales, Discount, Product, Price

Data OwnerVP, Merchandising

Data StewardPromotions Analyst

CDMPromotion

LDMpromotion: type, start_date, end_date, discount_pct, min_qty

PDMdim_promotion

ConsistencyAligned. Promotions are distinct from permanent price changes.

GlossaryPromotion

CDMPromotion

LDMpromotion

PDMdim_promotion

Order

Sales & Revenue High

A customer's expressed intent to purchase one or more products. An order may be pending, confirmed, fulfilled, or cancelled. Distinct from a completed sale.

Full Metadata

SynonymsSales Order, Purchase Order (from customer perspective)

Related TermsSale, Shipment, Return, Customer

Data OwnerHead of Sales Operations

CDMSalesOrder: distinct from SalesTransaction and Return

LDMsales_order: status ENUM (Pending|Confirmed|Shipped|Delivered|Cancelled)

PDMfact_order: grain is one row per order line item

ConsistencyAligned. Orders, Sales, and Returns are always modeled separately.

GlossaryOrder

CDMSalesOrder

LDMsales_order

PDMfact_order

Shipment

Logistics Medium

A physical movement of goods from a warehouse or distribution center to a customer delivery address or store location, associated with one or more orders.

Full Metadata

SynonymsDelivery, Dispatch, Consignment

Related TermsOrder, Carrier, Tracking, Return

Data OwnerVP, Logistics & Distribution

CDMShipment

LDMshipment: carrier_code, tracking_number, ship_date, delivery_date

PDMfact_shipment

GlossaryShipment

CDMShipment

LDMshipment

PDMfact_shipment

Return

Sales & Revenue Medium

A customer-initiated reversal of a completed sale, resulting in the return of goods and issuance of a refund or credit to the customer.

Full Metadata

SynonymsRefund, Credit, Reverse Sale

Related TermsSale, Net Sales, Order, Inventory

Data OwnerHead of Sales Operations

CDMReturn: modeled separately from SalesTransaction

LDMreturn_transaction: reason_code, refund_amount, return_date

PDMfact_return: linked to original fact_sales via original_order_sk

GlossaryReturn

CDMReturn

LDMreturn_transaction

PDMfact_return

Feature Store Record

Data Science High

A pre-computed, versioned, point-in-time correct feature vector derived from governed data products, stored in the centralized ML feature store for consumption by training and inference pipelines.

Full Metadata

SynonymsFeature Vector, ML Feature, Feature Row

Related TermsModel Training Dataset, Data Product, Customer, Product

Data OwnerHead of Data Science

Governing PolicyEvery feature must have: name, description, source lineage, owner, SLA. Versioned. PII excluded.

CDMAIMLFeatureSet

LDMfeature_record: feature_name, feature_value, entity_key, event_timestamp, version

PDMml_feature_store: partitioned by entity_type and event_date

GlossaryFeature Store Record

CDMAIMLFeatureSet

LDMfeature_record

PDMml_feature_store

Model Training Dataset

Data Science High

A point-in-time correct, labeled dataset assembled from the Feature Store for the purpose of training or retraining a specific ML model. Immutable once created, with full lineage to source data products.

Full Metadata

SynonymsTraining Set, ML Dataset, Training Snapshot

Related TermsFeature Store Record, Data Product, Model

Data OwnerHead of Data Science

Governing PolicyMust be immutable, versioned, PII-free, with documented feature set and label definition. Reproducible.

CDMDerived from AIMLFeatureSet

LDMtraining_dataset: dataset_id, model_name, created_at, feature_version, row_count

PDMml_training_snapshot: immutable, partitioned by model_name

GlossaryTraining Dataset

CDMDerived

LDMtraining_dataset

PDMml_training_snapshot

3 Enterprise Data Models

The CDM, LDM, and PDM are consistent with the business glossary. Every entity can be traced back to a glossary term.

Conceptual Data Model

11 business entities with glossary-aligned names, ontology relationships, and domain ownership.

Customer

Person/org that purchases. Domain: Customer Mgmt. Relationships: places Order, receives Shipment.

Product

Sellable item by SKU. Domain: Merchandising. Rel: BELONGS-TO Category, supplied by Supplier.

Store / Channel

Physical or digital sales location. Domain: Store Ops. Rel: PART-OF Region.

Supplier

External vendor. Domain: Procurement. Rel: PROVIDES Product.

Inventory

Stock at location. Domain: Supply Chain. Rel: Product AT Store.

SalesOrder

Customer intent to purchase. Domain: Sales. Rel: placed by Customer, contains Product.

SalesTransaction

Completed purchase. Domain: Sales. Rel: fulfills Order.

Promotion

Time-bound pricing incentive. Domain: Merchandising.

Shipment

Physical goods movement. Domain: Logistics.

Return

Sale reversal. Domain: Sales. Rel: reverses SalesTransaction.

AIML FeatureSet

Governed ML features. Domain: Data Science. Rel: derived from all entities.

Logical Data Model (Key Entities)

Customer: 9 attributes

Attribute	Type	Constraint	Glossary Link
`customer_id`	UUID	PK, NOT NULL	Customer → identifier
`full_name`	VARCHAR(200)	NOT NULL	Customer → name (PII)
`email`	VARCHAR(255)	UNIQUE	Customer → contact (PII)
`customer_type`	ENUM(B2C,B2B,Wholesale)	NOT NULL	Customer → type
`loyalty_tier`	ENUM(Bronze,Silver,Gold,Plat)	NULLABLE	Customer → loyalty
`registration_date`	DATE	NOT NULL	Customer → registration
`preferred_channel`	ENUM(Store,Online,App)	NULLABLE	Customer → channel pref
`ltv_segment`	ENUM(High,Med,Low)	NULLABLE	Customer → LTV
`primary_store_id`	UUID	FK → Store	Customer → primary store

Supplier: 10 attributes

Attribute	Type	Constraint	Glossary Link
`supplier_id`	UUID	PK	Supplier → identifier
`supplier_name`	VARCHAR(200)	NOT NULL	Supplier → name
`supplier_type`	ENUM	NOT NULL	Supplier → type
`country_code`	CHAR(3)	NOT NULL	Supplier → country (ISO 3166)
`lead_time_days`	INTEGER	NOT NULL	Supplier → Lead Time
`quality_rating`	DECIMAL(3,2)	CHECK 0–5	Supplier → Quality Rating
`contract_start_date`	DATE	NOT NULL	Supplier → contract
`payment_terms`	VARCHAR(20)	NOT NULL	Supplier → payment terms
`is_certified`	BOOLEAN	DEFAULT FALSE	Supplier → certification
`primary_category_id`	UUID	FK → Category	Supplier → category

Physical Data Model (Star Schema)

Governed naming convention: dim_ for dimensions, fact_ for facts, ref_ for reference, ml_ for ML objects. Snake_case, full words, no abbreviations.

dim_customer (SCD Type 2)

CREATE TABLE dim_customer ( customer_sk BIGINT IDENTITY PRIMARY KEY, customer_id UUID NOT NULL, -- glossary: Customer identifier full_name VARCHAR(200), -- PII: masked in non-prod email VARCHAR(255), -- PII: consent required customer_type VARCHAR(20) NOT NULL, -- glossary: Customer Type loyalty_tier VARCHAR(20), is_active_flag BOOLEAN, -- glossary: Active Customer registration_date DATE, effective_from TIMESTAMP NOT NULL, effective_to TIMESTAMP, is_current BOOLEAN DEFAULT TRUE );

fact_sales, dim_product, dim_supplier, dim_promotion DDL

fact_sales

CREATE TABLE fact_sales ( sales_sk BIGINT IDENTITY PRIMARY KEY, customer_sk BIGINT REFERENCES dim_customer, product_sk BIGINT REFERENCES dim_product, store_sk BIGINT REFERENCES dim_store, promotion_sk BIGINT REFERENCES dim_promotion, order_date_sk INT REFERENCES dim_date, quantity DECIMAL(10,3), unit_price DECIMAL(10,2), line_total DECIMAL(12,2), -- glossary: Sale amount currency_code CHAR(3) DEFAULT 'USD' );

dim_product

CREATE TABLE dim_product ( product_sk BIGINT IDENTITY PRIMARY KEY, product_id UUID NOT NULL, sku VARCHAR(30) NOT NULL, -- glossary: Product SKU product_name VARCHAR(200) NOT NULL, category VARCHAR(100), department VARCHAR(100), supplier_sk BIGINT REFERENCES dim_supplier, is_current BOOLEAN DEFAULT TRUE, effective_from TIMESTAMP, effective_to TIMESTAMP );

4 Glossary-to-Model Alignment Flow

In Scenario 1, every glossary term traces cleanly through CDM, LDM, and PDM. The flow is unbroken.

Alignment Flow: "Customer"

Business Glossary

Customer
"A person or org that has purchased from ShaDey Inc."

CDM

Customer
Entity with domain ownership, ontology

LDM

customer
9 attributes, business rules, quality rules

PDM

dim_customer
SCD2, surrogate keys, partitioned

Alignment Flow: "Net Sales"

Business Glossary

Net Sales
"Gross sales − returns − discounts"

CDM

Derived metric
SalesTransaction + Return + Promotion

LDM

Calculated field
Documented formula, single definition

PDM

vw_net_sales
View applying the single formula

When every layer uses the same definition, the same formula, and traceable naming, any report, dashboard, or ML model that consumes the data produces consistent, trustworthy results. There is one truth, and everyone can verify it end-to-end.

5 Term Deep Dives: Full Traceability

Tracing 6 key terms through every layer to demonstrate end-to-end consistency.

Glossary Term	CDM Entity	LDM Entity / Attribute	PDM Table / Column	Status
Customer	`Customer` (same name, same definition)	`customer`: 9 validated attributes	`dim_customer`: SCD2 with `customer_sk`	Aligned
Active Customer	Derived: Customer + SalesTransaction (12-mo rule)	`customer.is_active` (BOOLEAN, nightly recalc)	`dim_customer.is_active_flag`	Aligned
Product	`Product`: IS-A MaterialItem, HAS-MANY Variants	`product`: 8 attributes, SKU format rule	`dim_product`: SCD2, clustered on `sku`	Aligned
Net Sales	Derived: SalesTransaction − Return − Promotion discounts	Calculated field with documented formula	`vw_net_sales`: single enterprise formula	Aligned
Inventory On Hand	`Inventory`: quantity at location, excludes reserved	`inventory.quantity_on_hand` (INT, ≥ 0)	`fact_inventory_snapshot.quantity_on_hand`	Aligned
Supplier	`Supplier`: PROVIDES Product, HAS-A Contract	`supplier`: 10 attributes, quality_rating	`dim_supplier`: SCD2	Aligned
Promotion	`Promotion`: time-bound incentive	`promotion`: type, dates, discount_pct	`dim_promotion`	Aligned
Feature Store Record	`AIMLFeatureSet`	`feature_record`: name, value, entity_key, version	`ml_feature_store`: partitioned by entity_type	Aligned

In Scenario 1, every column in the PDM can be traced back to a glossary term via the LDM and CDM. Impact analysis for schema changes is straightforward: change the glossary term's definition, and you know exactly which tables, columns, and downstream consumers are affected.

6 Business Impact

How aligned governance and a consistent business glossary impact every aspect of data operations.

Analytics & Reporting

"Net Sales" means the same thing in every dashboard
Finance, Marketing, and Ops reports produce identical numbers
No reconciliation needed between team-specific definitions

Trust & Confidence

Business users trust the data because they defined the terms
No "which number is right?" debates in leadership meetings
Data products are certified and discoverable

AI/ML Readiness

Feature Store is semantically consistent with business glossary
Training datasets are reproducible and PII-compliant
Feature engineering takes 1–2 days, not weeks

Governance & Compliance

Full column-level lineage from glossary to PDM
PII identified at CDM, masked at PDM, excluded from ML
GDPR and SOX audit trails are complete

Interoperability

Cross-domain queries work because naming is consistent
New teams onboard quickly using the glossary as a guide
API contracts align with glossary definitions

Data Product Discoverability

Data catalog entries link to glossary terms automatically
Search for "Customer" finds all related tables, views, features
Self-service analytics enabled by semantic consistency

7 Key Takeaways

A centralized business glossary is the single source of truth. When CDM, LDM, and PDM all derive from it, the enterprise speaks one language.

Every column in the physical layer traces back to a business term. This makes impact analysis, compliance audits, and data quality investigations straightforward.

Semantic consistency at the modeling layer is the foundation for trustworthy AI/ML. Features derived from well-defined, governed data products produce reliable models.

Investing in glossary-to-model alignment pays dividends in every subsequent project: faster onboarding, faster feature engineering, fewer reconciliation errors, lower compliance risk.

Now see what happens when this alignment breaks down: Scenario 2: Misaligned Governance