← BIM, CAD & Digital Design Studio
Concept Explainer · BIM & CAD

Native BIM Formats vs. IFC

Why "opening the file" isn't the same as "getting all the data" — and why that gap has quietly derailed real coordination workflows.

A model saved as a native .rvtfile and the same model exported to IFC will both open, both show geometry, and both look — at a glance — like the same building. That surface-level similarity is exactly what makes the difference dangerous. One of these files is a vendor's complete internal representation of everything the authoring software understands about the model. The other is a deliberately generic, standardized translation built so other software can read it at all. Assuming those two things carry equivalent information is one of the most consequential and least-questioned assumptions in multi-platform BIM coordination.

The Setup

Two formats built for opposite jobs

A native format — Revit's .rvt, ArchiCAD's .pln, Tekla's .db1— is a vendor's own proprietary file structure, built to store every piece of information that vendor's own software understands: parametric relationships between elements, vendor-specific family and type behavior, proprietary metadata, and the full editing intelligence the authoring application relies on to let you keep modifying the model intelligently. It is complete by design — for that one application. Open it in a different platform and, in almost every case, you simply can't; the format was never meant to be portable.

IFC (Industry Foundation Classes)exists to solve exactly that portability problem. It is an open, vendor-neutral schema, published and maintained by buildingSMART, specifically designed so a model can move between Revit, ArchiCAD, Tekla, and dozens of other platforms without every party needing to own the same authoring software. IFC represents building elements — walls, doors, structural members, MEP equipment — using standardized, published definitions and property sets that any IFC-compliant application knows how to read. That's a genuinely valuable, genuinely necessary capability. It is also, unavoidably, a generic representation — and generic representations cannot losslessly encode every vendor-specific idea a proprietary format was built to store.

Native format (.rvt, .pln, .db1…)
Owned by: a single software vendor
Stores: everything that vendor's app understands
Fully readable by: that same vendor's software only
IFC
Owned by: buildingSMART — open, vendor-neutral
Stores: a standardized subset, per the published schema
Fully readable by: any IFC-compliant platform — with varying completeness

What survives an export to IFC

Not lossless
NATIVE FILE (.rvt)GeometryParametric family relationshipsProprietary metadataVendor-specific behaviorFully readable — only insidethe originating vendor's softwareExport toIFCIFC EXPORTGeometryStandard property sets(IFC Psets — carried over)Proprietary metadataVendor-specific behaviorLost or simplifiedin translation

Geometry and the standardized IFC property sets carry over reliably. Deeper, vendor-specific layers — the exact parametric logic behind a family, proprietary object behavior, non-standard metadata — don't have a defined home in the IFC schema, so exporters either drop them or flatten them into generic, less-intelligent geometry and text properties.

The same IFC file, opened in two platforms

Import quality varies
ONE IFC FILE(single standardized export)PLATFORM A IMPORT~90% of Psets mapped correctlyWalls, doors & MEP typed correctly,most properties readable nativelyPLATFORM B IMPORT~55% of Psets mapped correctlySome elements import as generic"proxy" objects, properties flattened

The IFC schema is the same for both. How completely and accurately each platform's importer maps that schema onto its own internal object model is a separate engineering effort per vendor — and it is not equal across platforms, or even across versions of the same platform. This is why the same IFC export can look faithful in one tool and noticeably degraded in another. (Percentages are illustrative — real completeness depends on the specific software versions and IFC MVD in use.)

Why this works

IFC trades some information for the ability to be read by anyone.

A native format doesn't need to compromise on what it stores — it only ever has to satisfy one application, so it can be as vendor-specific and information-dense as that application needs. IFC has the opposite constraint: to be usable by dozens of unrelated platforms, its schema has to describe building elements in terms every one of those platforms can agree on — standardized entity types (IfcWall, IfcDoor, IfcDistributionElement) and published property sets. That generality is the entire point of IFC, and it's genuinely valuable — but generality and vendor-specific completeness are in tension. A concept that only exists inside one vendor's parametric engine, with no IFC-schema equivalent, has nowhere standardized to go. Exporters make a judgment call: drop it, or approximate it as something more generic. Either way, the native file and its IFC export are no longer informationally identical, even though both still "open."

Common misconception
"Exporting to IFC and reopening it — even in the same software — preserves the model completely, with no data loss."

False, and it's a genuinely important coordination-workflow risk, not a theoretical one. IFC is a standardized, vendor-neutral schema — it was never designed to capture every proprietary parametric relationship, every vendor-specific family behavior, or every piece of software-specific metadata that a native format stores, because doing so would defeat the purpose of being vendor-neutral in the first place. Some information is genuinely lost or simplified on export, which is exactly why re-opening an IFC export — even back in the very same authoring application that created the original native file — can show a model that behaves or reports data differently than the native original. On top of that, real-world IFC export and import quality is not uniform: different software implements the IFC schema with different completeness and accuracy, and that completeness can change between versions of the same product. Coordination workflows that depend on IFC exchange between different BIM platforms should explicitly verify that the properties and relationships that actually matter — not just visual geometry — survived the round-trip, rather than assuming an IFC export is a perfect, lossless mirror of the native model.

Related Concept Explainers
LOD: Level of Development vs. Level of Detail
Read it →
Hard, Soft & Workflow Clashes
Read it →

Native BIM Formats vs. IFC — Concept Explainer

Explains why a native BIM file (such as Revit's .rvt) and its IFC export are not informationally equivalent, even though both open and both display the same building — using a layered-data comparison of what survives an IFC export and a side-by-side look at why the same IFC file imports with different completeness in different software.

Why This Is Commonly Misunderstood

Because both a native file and its IFC export can be opened and both show what looks like the same 3D model, it's an easy — and wrong — leap to assume they carry the same information. A native format like .rvt is a software vendor's own proprietary structure, built to store everything that vendor's application understands: parametric family relationships, proprietary object behavior, and vendor-specific metadata, in addition to geometry. IFC (Industry Foundation Classes) is a completely different kind of thing — an open, vendor-neutral schema published by buildingSMART, purpose-built so a model can move between unrelated BIM platforms (Revit, ArchiCAD, Tekla, and others) without everyone owning the same software. That neutrality is achieved by describing building elements in standardized, published terms every compliant platform can agree on — which is a different, and necessarily smaller, set of concepts than what any one vendor's proprietary format can express.

What Actually Gets Lost in Translation

Exporting to IFC is a real translation, not a lossless copy. Geometry and the properties covered by IFC's standardized property sets (Psets) generally carry over. But parametric relationships specific to one vendor's modeling engine, proprietary family/type behavior, and non-standard metadata don't necessarily have a defined equivalent in the IFC schema — so exporters either drop that information or approximate it as more generic geometry and flat text properties. This is exactly why reopening an IFC export, even in the very same software that created the original native file, can show differences from the original: some of what made the native model "intelligent" simply has nowhere standardized to go in IFC.

Where This Matters — IFC Compliance Quality Varies Between Vendors

The IFC schema is the same for every compliant platform, but how completely and accurately a given piece of software implements IFC import and export is a separate engineering effort per vendor — and it is not uniform. Real-world IFC mapping quality varies meaningfully between different BIM platforms, and even between versions of the same platform, as vendors improve (or occasionally regress) their IFC translators over time. Coordination workflows that pass models between different BIM platforms via IFC — a common practice in multi-discipline projects — should treat IFC export as a translation to verify, not a guaranteed lossless mirror. Confirming that critical properties, quantities, and relationships (not just visual geometry) survived the round-trip is a standard, necessary QA step, particularly for handoffs feeding COBie deliverables, quantity takeoffs, or downstream facility-management data.

Frequently asked questions

If I export a Revit model to IFC and reopen it in Revit, will it look exactly the same?

Visually, usually very close. Informationally, not necessarily identical. Some Revit-specific parametric relationships, family/type behavior, and proprietary metadata don't have a defined equivalent in the IFC schema, so they can be dropped or simplified on export — meaning the reopened IFC version can behave or report data differently than the original native .rvt file, even inside the same software.

Is IFC a bad format, then?

No — it solves a problem native formats fundamentally cannot: letting different BIM platforms exchange models without everyone owning the same software. That is a real and necessary capability. The point is not that IFC is flawed, but that its vendor-neutral design means it cannot capture every proprietary detail a native format stores, so it should be treated as a standardized exchange format with known limits, not a lossless mirror.

Why would the same IFC file look more complete in one BIM platform than another?

Because each vendor writes its own IFC import/export translator, and the completeness and accuracy of that translator — how well it maps IFC entities and property sets onto that platform's internal object model — varies by vendor and by software version. The IFC schema itself doesn't change between platforms; how well each platform implements it does.

How should a coordination workflow protect against IFC data loss?

Explicitly verify what actually needs to survive the round-trip — critical property sets, quantities, classifications, and relationships relevant to the deliverable (e.g., COBie data, MEP system connectivity) — rather than assuming visual geometry matching means the exchange was complete. Many BIM Execution Plans specify which IFC Model View Definition (MVD) and property sets are required for a given exchange precisely because of this variability.

🎓

Try our BIM, CAD & Digital Design Studio

More calculators, simulators, and guides for this discipline.

Related tools & guides

BIM vs. CAD: What's the Difference and Why It MattersAutodesk Revit Basics: UI, File Types & the Parametric AdvantageBIM, CAD & Digital Design System ArchitectureBIM Execution Plan Builder