Why Data Modelling Training & Certification Matters with Hans Hultgren

Executive Summary

This webinar highlights the key aspects of the Data Vault journey in enterprise data warehousing, addressing challenges such as data silos through unified decomposition and the Key Dependency Principle. Hans Hultgren emphasises the importance of agility and incremental building when working with dimensions, along with Effective Lifecycle Management (ELM) for both business users and technical automation. Additionally, he discusses the significance of data quality, master data management, and self-service business intelligence, and offers insights into certification programs and resources to get started on this transformative path.

Webinar Details

Title: Is Your Data GenAI Ready for Data Professionals with Paul Grobler
Date: 2026-02-12
Presenter: Howard Diesel & Paul Grobler
Meetup Group: African Data Management Community
Write-up Author: Howard Diesel

Introduction & The Data Vault Journey

The webinar opens with a reflection on the evolution of Data Vault, tracing its origins back nearly two decades. Hans Hultgren discusses his early work, including his first book on Data Vault, and how the methodology surprised many with its broad applications—from master data management to data quality.

The conversation between Howard and Hans highlights the emergence of Ensemble Logical Modelling (ELM), a pattern that extends beyond just Data Vault to represent a broader approach to modelling business concepts. Hans explains that finding the right name for this pattern took months because “all modelling names were already taken,” but eventually, the team settled on ELM to describe the logical grouping of business concepts.

This approach has matured significantly over the last 15 years, with Remco Broekmans recently publishing a book dedicated to ELM. The presenters note that this modelling direction aligns perfectly with the current technological landscape, particularly with the rise of AI and code automation, which benefit from the structured, pattern-based nature of Data Vault.

Figure 1 ‘About Data Vault Modeling’

Figure 2 DWBI: the Issues and the Requirements

Figure 3 Common Issues with Enterprise Data Initiatives

Core Challenges in Enterprise Data Warehousing

Hans outlines the persistent issues plaguing enterprise data warehousing: the need for agility, business focus, incremental build capabilities, and full auditability. He contrasts the two traditional dominant approaches: Bill Inmon’s third normal form (3NF) and Ralph Kimball’s dimensional modelling.

While Inmon’s approach offered an enterprise view, it often resulted in long development cycles and complex re-engineering. Conversely, Kimball’s dimensional approach provided speed and responsiveness but frequently led to data silos and anomalies when enterprise-wide integration was attempted. Additionally, Hans emphasises that neither traditional approach fully addresses the unique requirements of a modern data warehouse, which must integrate data from diverse sources while remaining adaptable to change.

The hardening of architecture in both 3NF and dimensional models makes them difficult to modify once built. The solution proposed is Unified Decomposition—a method that splits data into smaller, manageable parts without losing the cohesive business meaning, effectively addressing the challenges of re-engineering costs and lack of agility.

Figure 4 Need Modelling Pattern Optimised for Enterprise Data

Understanding Unified Decomposition – The Core Pattern

At the heart of Data Vault is the concept of Unified Decomposition, which Hans describes as an oxymoron that makes perfect sense for agility. To avoid the re-engineering penalties of traditional models, Data Vault splits business concepts into three distinct components: Hubs (representing unique business keys), Links (representing natural business relationships), and Satellites (containing all context, descriptives, and history).

Together, these components form an Ensemble—a logical grouping that serves as a single business concept (such as “Customer” or “Product”). The critical technical principle here is maintaining key dependency. Just as in 3NF or dimensional modelling, every attribute must depend on its primary key. By decomposing the model, Data Vault allows organisations to add new context or relationships without disrupting existing structures. This separation enables the warehouse to grow incrementally while maintaining the rigorous data integrity found in traditional relational models.

Figure 5 DWBI: the Solution in a Nut Shell

Figure 6 Unified Decomposition

Figure 7 The Data Vault Ensemble

Figure 8 Ensemble

Figure 9 CBC Instance: Ensemble Identifier / Key / Hub

Figure 10 NBR Relationship: Association / Link

The Key Dependency Principle & Avoiding Data Silos

Hans moves on to explore the most critical rule of Data Vault modelling: maintaining one-to-one relationships in Satellites. He explains that a Satellite must relate to its Hub (or Link) in a strict one-to-one manner for any given point in time, identical to how attributes relate to a primary key in a 3NF entity or a Type 2 dimension. If this rule is broken, the model loses its integrity, leading to data anomalies and silos.

In the context of data migration and integration, when dealing with multiple business units or legacy systems with different referential integrity constraints, Data Vault shines by allowing alignment rather than forced integration. Instead of forcing diverse business units to agree on a single definition of “Customer,” Data Vault can accommodate nuanced definitions side by side. Hans admits this approach initially feels like “petting the cat the wrong way” for experienced modellers, as it requires unlearning habits from operational system modelling, but it ultimately provides the flexibility required for true enterprise integration.

Figure 11 NBR Relationship: Association / Link

Figure 12 Satellite

Figure 13 “Data Vault means thinking Differently”

Figure 14 “Data Vault means thinking Differently” Continued

Figure 15 Data Vault Example

Figure 16 Data Vault Example pt.2

Figure 17 “Data Vault means Thinking Differently”

Learning Data Vault – The Educational Journey

Hans famously states that “Data Vault modelling is easy,” but quickly qualifies that it requires a significant paradigm shift. The mechanical act of creating Hubs, Links, and Satellites is straightforward, but understanding why and when to use them requires unlearning traditional patterns.

The Genesee Academy’s training approach reflects this, offering a comprehensive three-day instructor-led course preceded by two weeks of video lessons (over 60 videos) to prepare students. The certification exam is rigorous, with a 68% first-time pass rate, emphasising that it validates genuine mastery rather than just attendance.

Hans stresses that you cannot learn Data Vault solely from a book or by asking ChatGPT; it requires experiential learning through interactive cases and exercises. Students must work through the confusion to reach the “aha moments” where the pattern clicks. Once understood, the logic becomes intuitive, and the methodology’s benefits become undeniable.

Figure 18 Modelling New Subject Areas

Figure 19 Data Vault Modelling

Figure 20 Learning Data Vault Modelling

Figure 21 The Certified Data Vault Data Modeller (CDVDM)

Figure 22 The Certified Data Vault Data Modeller (CDVDM) pt.2

Figure 23 CDVDM Certification Exam

Figure 24 Summary with Q&A

Agility, Incremental Build & Working with Dimensions

A major advantage of Data Vault is its support for incremental delivery. When a new source system or subject area is introduced, the existing model doesn’t need to be re-engineered. Instead, new Links and Satellites are simply “plugged in” to the existing Hubs. This additive nature allows the data warehouse to evolve organically without breaking existing reports or ETL processes.

Hans also clarifies, upon request, the relationship between Data Vault and dimensional models. While Data Vault is the storage layer, data is typically presented to users via dimensional marts. Hans explains that creating a dimension is as simple as rolling up a Hub key with selected attributes from various Satellites.

Interestingly, while a “Customer” concept might have 168 attributes in the warehouse, a specific business report typically needs only 2-5 of them. This allows for the rapid creation of purpose-built dimensions derived directly from the resilient Data Vault core.

ELM for Business Users & Technical Automation

Ensemble Logical Modelling (ELM) workshops are designed to bridge the gap between IT and business. Unlike technical data modelling sessions that can alienate business users, ELM workshops focus on identifying core business concepts and their natural relationships using a node-edge approach. This resonates with business stakeholders because it mirrors how they think about their operations—customers buy products, employees work at stores, etc.

From a technical architecture perspective, an attendee adds a crucial insight: the simplicity of the Data Vault pattern (Hub, Link, Sat) enables powerful automation. Instead of writing thousands of unique SQL scripts (“blobs of code”), the consistent pattern allows teams to automate the generation of loading code based on metadata. This means the technical team can focus on capturing context and business rules, while the repetitive task of moving data into the warehouse is automated, delivering massive efficiency gains.

Figure 25 Links and Information

Data Quality, Master Data Management & Self-Service BI

Data Vault plays a unique role in Master Data Management (MDM) and data quality. By ingesting raw data from all sources without premature integration, the Data Vault acts as a mirror of the business, revealing data quality issues (such as two customers on a single sale) that operational systems might hide.

It allows organisations to see the data exactly as it arrives, providing a foundation for creating “gold records” downstream without losing the original context. This structure is essential for enabling true self-service BI. Hans explains that self-service often fails because users don’t understand the semantic transformations that happen in traditional warehouses.

By keeping the data model aligned with business concepts from source to target, Data Vault preserves the business terminology users understand. This creates a fluid “digital” path for data, ensuring that when a business user queries “Loans,” they receive data that aligns with their operational understanding, minimising semantic errors and building trust in the data.

Certification, Resources & Getting Started

The webinar concludes with practical steps for those looking to master Data Vault. Upcoming in-person training is highlighted for May 11-13 in Cape Town, offering a unique opportunity for immersive learning and networking. For those unable to travel, online options are available, though the presenters emphasise the value of the in-person experience. All students receive access to pre-course video lessons two weeks in advance to ensure they are ready for the deep dive.

Resources such as dbstandards.com and elmstandards.com are recommended for further reading. The team at Genesee Academy (Hans and Remco) encourages engagement and offers support to past students who may need a refresher. They advise that while the core warehouse team needs deep technical certification, broader business teams can benefit from understanding the high-level ELM concepts to participate effectively in modelling workshops.

Figure 26 Upcoming 3-day Certified Data Vault Data Modeller classes