The 3 Levels of BIM Quality Checks: How Building Owners Tune the Algorithm

What does quality mean for a property? For some it is shadow-free walls, for others premium materials. In the digital age, there is a far more pragmatic definition: quality is when the delivered model exactly meets the client's requirements.

The problem: most building owners only know what they want when they see it. In early concept phases that is fine—you need flexibility and iteration. But once execution begins, planning changes cost serious money.

To minimize this risk, Swiss building owners usually choose two classic strategies:

• More staff: Hire more owner representatives than there are planners on the project.
• More paper: Commission 3,000-page specification documents to legally cover every eventuality.

In times of acute skills shortages and digital transformation, that is the wrong path. It is time for strategy number three: we must tune the algorithm. An automated check routine does not care how often it repeats the work. It does not get tired.

Risk Management Through Automated Model Checks

A real example: during planning, building services ducts turn out thicker than expected. Walls must be rebuilt wider. Bathrooms shrink by a few centimeters unnoticed. The result? Planned sanitary fixtures no longer fit, the marketing concept wobbles, the business case breaks down.

Whoever notices this error on site pays an expensive lesson. Whoever checks the model automatically catches the error before it happens. But how do you build such a QA system?

We divide BIM quality verification—by analogy with software development—into three logical levels:

[Level 3: System Test]        → Complex business rules & simulations
        ↓
        [Level 2: Integration Test] → Geometric logic & modeling rules
        ↓
        [Level 1: Unit Test]          → Pure data & attribute presence

Level 1: The Unit Test (Pure Data Presence)

At the lowest level, we verify whether required data exists at all.

• The goal: Query gross floor area (GFA) or net living area.
• The requirement: Every room element must be modeled as IfcSpace with a correct PredefinedType (e.g. GFA or SPACE).
• The conclusion: This is the absolute minimum. It only tells us whether planners filled in the table—not yet whether the content makes sense.

Level 2: The Integration Test (Modeling Logic)

Here we verify how things were modeled to catch complex geometric risks.

• The goal: Ensure the underground garage has enough parking spaces that are physically drivable.
• The requirement: Each parking space must not be a lump-sum area—it must be modeled as an individual IfcSpace element.
• The check: The algorithm does not only count parking spaces—it automatically verifies clear height, width, and whether the access lane meets the legal minimum turning radius.

Level 3: The System Test (Business & Authority Rules)

The top tier. Here we link data from levels 1 and 2 with external rules (e.g. building law or investor requirements).

• The goal: Does the unit mix and parking ratio work for the entire site?
• The requirement: Individual rooms (IfcSpace) must be grouped via zones or systems into a specific apartment (e.g. "Apartment 3.1, 4.5 rooms").
• The check: The algorithm dynamically calculates the number of apartments, compares it to parking spaces, and fully automatically verifies whether the project meets municipal regulatory requirements.

Conclusion: Digitizing Expert Knowledge

The true value of BIM for building owners is not at level 1. Anyone can read Excel lists. Profit is created at levels 2 and 3.

By defining these check rules cleanly once, you pour your best project managers' internal know-how into digital tools. That relieves your teams from dull control work, secures your projects, and ensures you get exactly the quality you ordered.

Stop checking data manually—contact pragmaticBIM to set up your automated QA algorithm