[2026 Latest] Decentralized Data Governance and Silo Elimination via Data Mesh

"We were supposed to have integrated all company-wide data, yet it's not being used at all on the front lines" or "The IT department has become a bottleneck for data extraction"—this is the reality many companies face. The true nature of this challenge lies in the limitations of "centralized architectures" that rushed physical data consolidation while pushing management responsibility solely onto a central authority. In 2026, the key to achieving true data-driven management lies in transitioning to a "Data Mesh," where each department autonomously manages and provides data. In this article, we will delve into strategies for building a modern data infrastructure that breaks down silos and turns company-wide data into value, from a practitioner's perspective.

1. The Limitations of "Organizational Silos" in Centralized DWHs

A common scenario we see in our consulting work is where hundreds of millions of yen are invested to build data lakes or DWHs (Data Warehouses), but the only people who understand the actual content of the data are the "departmental staff." The IT department cannot interpret the meaning of the data, resulting in a pile-up of data that no one can use. This is not a technical silo, but a negative consequence created by "responsibility silos."

In traditional centralized models, a disconnect existed where data was generated on the front lines but managed by the IT department. However, in today's accelerating business environment, it is impossible for a central data team to process every data request. In fact, there is a significant gap in the "lead time for data delivery" between companies that are successfully utilizing data and those that are stagnant.

2. Data Mesh: Domain-Driven Autonomous Decentralized Governance

Data Mesh is a concept that defines data not as an "asset to be managed centrally," but as a "product provided by each department (domain)." For example, the marketing department is responsible for publishing "customer behavior data products" and the logistics department for "inventory trend data products" in a format that is easy for other departments to use.

In our actual consulting projects, we spend the most time instilling this concept of "data products." The first step to breaking down silos is not simply enabling SQL queries, but defining "for whom, for what purpose, and at what level of quality" that data is provided. In this process, the "user-centric UI/UX design" philosophy cultivated in our D2C EC Site Construction & Growth Support services functions very effectively in organizing data catalogs.

3. Four Steps to Building a Modern Data Infrastructure

To build a modern data infrastructure and turn company-wide data into value, the following steps are essential:

• Domain Identification: Clarify which department should hold "sovereignty" over which data.
• Provision of Self-Service Infrastructure: Establish platforms (such as Snowflake or BigQuery) where each department can process and publish data without specialized technical knowledge.
• Federated Governance: Implement mechanisms to automatically apply company-wide rules (personal information protection, naming conventions).
• Data Product Evaluation: Turn the utilization of data into KPIs and run improvement cycles.

Of particular importance is the change in the role of the IT department. They must shift from being "data managers" to "platform providers" that enable each department to utilize data. A common failure on the front lines is introducing tools while leaving this role change ambiguous.

4. The "Data Ownership" Barrier Faced on the Front Lines

In the implementation of Data Mesh, a higher barrier than technology is the resistance from the front lines: "Why should our department have to take on the responsibility of data management?" In our actual support, to lower this psychological hurdle, we start by creating a "success story through data utilization" in a single domain as a small start.

For example, in D2C EC Site Construction & Growth Support, by showing concrete results—such as reducing wasted advertising costs due to stockouts by 30% through real-time linking of advertising and inventory data—we stimulate ownership in other departments. The essence of a data utilization strategy lies not in tool selection, but in this transformation of organizational culture.