When quality relies on documents, learning becomes too costly

Table of Contents

Picture of Ole Kristian Kvarsvik
Ole Kristian Kvarsvik
Japan demonstrates just how demanding it becomes when complex construction projects are still managed through documents, and why the next phase of BIM is all about structured data and learning.

A high-tech market with document-based processes

Japan is one of the markets that most clearly demonstrates how demanding it becomes when complex construction projects are still based on document-centric information flows.

It may seem paradoxical. Japan is associated with high quality, high discipline and advanced technology. Yet the construction industry remains heavily document-centric. Much is planned down to the smallest detail, and much of the communication to all downstream processes still takes place through drawings, specifications and other documents. This provides predictability. But it also comes at a cost. However, the cost is not just about time and efficiency.

It is also about people.

In a system where quality is ensured through ever-increasing numbers of documents, checks and manual processes, the burden on those actually carrying out the projects increases. In Japan, this has historically been managed through high staffing levels and extensive overtime. Today, the situation is different. The workforce is ageing, the supply of skilled personnel is lower, and new regulations restrict the use of overtime. As a result, document-based processes become not just inefficient. They become a real constraint on delivery capacity.

Data-centric BIM is the way forward in Japan
The Osaka Dojihama Tower, completed in 2024, was one of the first StreamBIM projects in Japan.

Why document-centric processes become a bottleneck

When information is moved out of the model and into documents, duplicates arise. This is also why document-centric processes create versioning issues and hinder learning in BIM projects. Changes must be updated in many places, and one loses track of which version is current. When quality, deviations and decisions are also documented in separate threads, the insight becomes fragmented. The project documents more, but does not necessarily learn better.

In many cases, deviations lead to new control routines and more documentation.

Over time, this builds up a comprehensive quality regime, but without the insights necessarily becoming more accessible or usable. When deviations, actions and decisions are stored in separate documents and systems, the ability to analyse across them is lost. The result is that the organisation becomes better at documenting, but not necessarily better at improving.

This is where Japan becomes interesting.

Not because the country is necessarily ahead in everything, but because the market highlights some questions that many others are also facing: How much quality assurance can a project tolerate before it becomes too cumbersome?

How does one learn across projects when the insights are locked away in documents? And how does one streamline operations when the workforce is becoming scarcer?

This is particularly relevant in Japan right now. The construction industry faces both recruitment challenges and a need for efficiency, whilst tolerance for errors is low. When quality requirements are high and there is little room for manoeuvre, the weaknesses in document-based processes become apparent more quickly.

It is no longer enough to simply document more. We must learn better.
The next phase of BIM is about data that can be used and learned from.

The next phase of BIM is not about a single change, but about two parallel shifts

  1. One concerns how information flows from design to the construction site.
  2. The other concerns what happens to the data created during execution – whether it becomes documentation, or a basis for learning and improvement.


It is particularly the latter that is now becoming crucial. That is why the next phase of BIM is not just about viewing the model. It is about turning the model and data into a shared, structured source of truth. When deviations, quality assurance and decisions are linked to the same dataset, quality becomes a basis for analysis and learning, not just documentation. This provides a better basis for identifying patterns, reducing recurring errors and improving work processes over time. It does not mean that documents will disappear. But it does mean that more processes can be transitioned to a more data-centric way of working, step by step.

At the same time, this is not a matter of replacing current working methods overnight

In markets such as Japan, drawings and documents will continue to form part of the production basis for many years to come. The question is how to take the next steps, and where to start. Experience shows that the greatest gains often lie in linking quality assurance, deviations and decisions directly to the model and a shared dataset. This is where we move from documentation to insight.

For StreamBIM, this is the core: helping the market move from document-driven complexity to a structured flow of information throughout the entire lifecycle.

When the model, deviations, quality and decisions are linked to the same data foundation, the ‘single source of truth’ becomes a practical reality, not just a slogan. It then becomes possible to combine better quality and less friction with stronger traceability, control and learning. It is precisely in this shift that StreamBIM becomes relevant as a platform.

When data is structured and linked to the model, it opens the door not only to better documentation. It provides the opportunity to analyse patterns, understand causes and, over time, predict where risks and deviations will arise.

This is what lies in the next step: not just building on truth, but analysing and eventually predicting it.

Our claim

Japan demonstrates not only how much quality can be built into projects. It also shows how costly it becomes when quality is supported by documents rather than structured data.
The platforms that will succeed in the future are not those that simply display the model best. They are those that make the data usable, traceable and capable of learning throughout the entire lifecycle.

A single source of truth is no longer just a vision. It is becoming a requirement.

Build on truth – analyse on truth – predict ‘the truth’

Do you want to reduce data breaks/data loss between project stages and get a better foundation for learning and quality assurance in your projects?

Talk to us about how StreamBIM is used in practice.