Home / Blog / Scan-to-BIM data checking

Blog · Scan-to-BIM · Quality

Scan-to-BIM data checking: how to validate a model against its point cloud

Scan-to-BIM: as-built BIM model checked against the laser scan point cloud
From scan to model: every delivery is checked back against the point cloud.

An as-built model is a promise: that the geometry on screen matches the building as surveyed. Scan-to-BIM data checking is how you verify that promise — before retrofit, quantities or facility decisions are taken on top of it. Here is what a serious check covers, and the three questions that reveal whether a provider actually runs one.

Why checking matters more than modelling

Modelling is the visible half of scan-to-BIM; checking is the half that decides whether the model is usable. A wall drawn 5 cm off its true position propagates into quantity take-offs, clash-free routing that clashes on site, and refurbishment designed around openings that are not where the drawings say. The scan-to-BIM services market keeps growing on retrofit and renovation demand (MarketIntelo 2025 estimates USD 6.2 bn in 2025 for scan-to-BIM services) — and every one of those projects starts from the same question: can I trust this geometry? A documented check answers it in writing, before the model enters your workflow.

The four checks that come before modelling

A proper quality gate runs on the scan data itself, before a single wall is modelled:

1. Registration, station by station

Individual scans are aligned into one coordinate frame; the residuals of that alignment are the first number to look at. High residuals in one area mean the cloud is weak exactly there — and anything modelled on top inherits the weakness.

2. Coverage of the agreed scope

The cloud must cover everything the scope lists, with no silent gaps. Uncaptured zones (locked rooms, unsafe areas, occlusions) are normal — what is not acceptable is discovering them after delivery. They belong in the report and in the model from day one.

3. Density against the agreed detail

Point density must support the level of development you ordered: fine mouldings at LOD 300 need denser data than a warehouse shell. Density is agreed up front, not debated after.

4. Naming, units and coordinate system

Files arrive with a naming convention, consistent units and a declared coordinate system, checked against your standards — or against UNI 11337 / ISO 19650 where no client standard applies. This is the unglamorous check that decides whether the model plugs into your CDE without a week of rework.

The deviation report: model against cloud, in writing

After modelling comes the second gate: the model is compared back against the point cloud, element by element. The output is a deviation report with measured distances, out-of-tolerance flags and the uncaptured zones declared explicitly. Ask for it with every delivery: it is the comparison you would run yourself, already done and documented. A provider who cannot produce one has not checked — whatever the brochure says.

Tolerances are conditional — anyone quoting a bare number is guessing

Achievable tolerance follows the instrument, the scanning distance, the point density and the complexity of the geometry. A bare “±X mm” without conditions is not a specification, it is a wish. State the tolerance with its conditions in writing, and set the checking level to what the model is for: retrofit detailing, quantity extraction and facility management do not need the same gate — and the strictest gate is not always the cheapest fit. If a use case only needs reliable room areas, paying for millimetre-grade validation of every moulding is waste, not quality.

Three questions for your next provider

Will I receive a deviation report of model against cloud with the delivery? Under which stated conditions does your tolerance hold — instrument, distance, density? And which zones could not be captured, and where are they declared? Three concrete answers, three pieces of paper. Everything else is marketing.

In short

Scan-to-BIM data checking is registration, coverage, density and standards before modelling — plus a written deviation report after it. Commission the checks, not just the model, and the as-built geometry becomes something you can build decisions on.