Introduction 

In today’s fast-paced marketing landscape, data is at the heart of every strategic decision. For global advertising and marketing leader Dentsu, managing large and complex semantic models built on top of dozens of source systems is critical to delivering actionable insights for their clients. As the company scales its analytics capabilities, ensuring performance, accuracy, and collaboration in Power BI semantic models becomes increasingly challenging. 

In addition to client-related advertising and marketing reporting, Dentsu has invested in an internal performance management reporting platform. It is crucial to monitor the performance of the multinational agency network from sales, financial and performance perspectives in order to enable data-driven decision-making. 

This article explores how Dentsu leveraged Tabular Editor to overcome these challenges, enabling faster development, better DAX code quality, optimized performance, and more efficient team collaboration on enterprise-grade semantic models. 

Dentsu’s challenges on scaling and optimizing the Enterprise Semantic Model  

Managing data at scale presents unique challenges, and for Dentsu, this was particularly evident in the design of their enterprise semantic models. With a large number of source systems feeding into shared fact tables, the team needed a way to reconcile and organize data efficiently. Although data was distributed across multiple Synapse tables, all of these ultimately had to flow into a single fact table at the semantic layer. To support both performance and manageability, Dentsu required the ability to create multiple partitions per fact table, including yearly partitions. However, with certain models requiring hundreds of partitions the idea of creating and maintaining them manually was not realistic. Such an approach would have been highly repetitive, error-prone, and consumed days of effort with every update cycle. 

The scale of data also introduced new performance demands. Dentsu’s semantic model between 50GB-100GB in size, meaning model refresh times could quickly balloon and threaten to exceed the available maintenance windows. A refresh that overruns its scheduled window risks delaying downstream processes and impacting reporting availability. The team needed a strategy to identify where bottlenecks were occurring, whether in DAX calculations, data compression, or storage design. Simply scaling up Power BI Premium capacity was not an efficient or cost-effective option, so optimization at the model level became essential. 

Beyond scale and performance, collaboration added another layer of complexity. With six developers working on the same semantic model at the same time, parallel development was challenging. Power BI Desktop alone offered limited support for team-based development, and without proper version control, there was no reliable way to track who made changes, understand what was modified, or revert to a stable version if something went wrong. This lack of visibility increased the risk of conflicts, overwrites, and rework. Additionally, the team needed a structured approach to deployment across multiple environments (development, test, and production). Without automation and governance, maintaining consistency across these environments would be difficult and prone to human error. 

At Dentsu, the adoption of Tabular Editor has been a game-changer in our journey toward building enterprise-grade Power BI semantic models that are not only performant and scalable but also sustainable and maintainable over time. - Patrick Sura, Director Business Intelligence, EMEA, Dentsu

How Dentsu solved complex challenges with Tabular Editor 

Partition Management at scale 

To manage large-scale data efficiently, Dentsu relies on Tabular Editor to streamline the creation of partitions within their semantic models. For a single model, the team manages hundreds of partitions. Manually creating and maintaining these would be a tedious, time-consuming task, taking hours or even days. By using C# scripting in Tabular Editor, Dentsu automates this process, completing the work in just minutes while ensuring accuracy and consistency. 

Model optimization for performance 

With a semantic model between 50GB-100GB, performance optimization is critical for Dentsu. While Power BI’s built-in Performance Analyzer highlights which visuals run slowly, it does not reveal the underlying cause of performance bottlenecks. By adopting Tabular Editor’s Best Practice Analyzer, VertiPaq Analyzer, and DAX Optimizer, Dentsu can dig deeper – identifying inefficient DAX expressions, storage engine issues, and modelling patterns that impact speed. This enables the team to fine-tune their semantic model and deliver significantly faster and more reliable performance at scale. 

Parallel development and DevOps enablement 

Collaboration is key when multiple developers contribute to the same semantic model. To support this, Dentsu uses Tabular Editor’s Save to Folder feature to break the model into metadata files, which are then managed through Git repositories. This approach allows the team to track changes, review version history, and revert updates when needed. By integrating this workflow into their DevOps pipeline, Dentsu has also automated deployments – making it possible to push the same model into three different workspaces (environments) with speed and consistency. This not only accelerates delivery but also strengthens governance and reliability. 

Boosting productivity building semantic models 

At Dentsu, the team leverages Tabular Editor to significantly improve productivity when building semantic models in Power BI. By harnessing the power of C# scripting, they streamline repetitive tasks such as batch renaming objects, updating M code to replace schemas in queries, and even generating model documentation automatically. In addition, they make use of Macros, which can be saved and rerun to accelerate routine workflows, ensuring consistency and efficiency across projects.  

Accelerating and enhancing DAX development 

Dentsu boosts the quality and speed of its DAX development by leveraging Tabular Editor’s advanced coding features. The Code Assist capability helps developers write DAX measures faster with intelligent suggestions that reduce syntax errors and accelerate coding. With Peek Definition, the team can instantly view the DAX code of referenced objects without navigating away from their current work, keeping context intact and saving time. The built-in DAX Formatter ensures code is always clean, consistent, and easy to read across the team. In addition, Code Actions proactively suggest improvements – such as following best practices, avoiding common pitfalls and anti-patterns, eliminating deprecated functions, and refactoring code for clarity.  

Leveraging advanced modelling features 

Dentsu takes full advantage of Tabular Editor’s advanced modelling capabilities to build and manage enterprise-scale semantic models more effectively. The team creates perspectives in Tabular Editor to create simplified, role-based views that make models easier for end-users to navigate. With the deploy model feature, they can quickly generate copies of models, streamlining testing and iteration. The diagram view allows developers to visually manage and create relationships between tables, while data refresh provides Dentsu the ability to preview data directly within Tabular Editor for faster validation. To keep models organized and maintainable, Dentsu also creates folders that group related objects logically.  

Tabular Editor as a cornerstone of Dentsu's BI strategy

By adopting Tabular Editor, Dentsu has transformed the way it builds, manages, and optimizes enterprise semantic models. Automation and scripting have streamlined partition management and routine model maintenance, while advanced optimization tools ensure high performance even at massive scale (50GB-100GB model refreshes several times per day). Developers can write cleaner, faster DAX, maintain organized models, and collaborate effectively across multiple environments – all without compromising governance or quality. For Dentsu, Tabular Editor has become a cornerstone of their business intelligence strategy, empowering the team to deliver faster insights, improve productivity, and support data-driven decision-making at a global scale.