Ask three vendors to price a scan to BIM package and you can get three numbers that differ by a factor of four. The instinct is to assume somebody is padding. Usually they are not — they are answering three different questions, because the brief left the expensive variables undefined.
Square footage is the unit everyone quotes in. It is almost never the thing that decides the price.
The five variables that actually move the number
1. Tolerance
This is the single largest cost driver, and the one most briefs omit entirely.
"Model the building" can mean within 50mm, good enough to plan a fit-out or within 3mm, good enough to prefabricate a facade panel against. The second requires denser scanning, more control points, more registration effort and a slower, more careful modelling pass. The geometry looks similar. The hours are not.
State a tolerance, and state what it applies to. Structure often needs to be tighter than finishes; a single project-wide figure priced at the tightest requirement means paying survey-grade rates for a suspended ceiling nobody will ever prefabricate against.
2. What you want modelled, element by element
A point cloud contains everything the scanner could see. A model contains only what somebody chose to build. That choice is the scope.
Structure, envelope and primary MEP runs are the common core. The cost escalates with:
- Small-bore services. Every branch, hanger and valve modelled individually is a large multiplier on MEP hours. Modelling above-ceiling services to 50mm and below is a very different job from modelling primary distribution only.
- Existing finishes and joinery. Rarely needed, frequently requested by default.
- Concealed elements. A scanner cannot see through a wall. Anything hidden must be inferred from drawings, opened up, or excluded — and the brief should say which.
3. Site access and occupancy
Scanning an empty shell is straightforward. Scanning an operating hospital, a live retail floor or a plant on a two-hour maintenance window is not.
Occupancy costs in two ways. The obvious one is time on site: out-of-hours working, escorts, permits, more setups to see around obstructions. The less obvious one is occlusion — furniture, stock and equipment block the scanner, leaving gaps that must be resolved in the model by inference rather than measurement. Gaps are slow, and they are where accuracy quietly degrades.
4. Geometric complexity
A rectilinear warehouse and a heritage building of the same floor area are not comparable work. Curved and warped surfaces, non-planar walls, ornate mouldings, and out-of-plumb structure all resist the parametric tools a modeller works with. Historic buildings are the classic case: nothing is straight, nothing repeats, and every element is bespoke.
Repetition is the cheapest thing in a model. Twenty identical floors cost far less than twenty floor area equivalents of unique geometry.
5. The deliverable, and the software it lands in
The model is rarely the end of the job. Price changes with:
- Native format vs IFC. A native Revit model with usable families is more work than an IFC export, and far more useful downstream.
- Level of Development. See LOD 300 vs LOD 400 — the distinction is worth settling before anyone quotes.
- Drawings off the model. Plans, sections and elevations are a separate production effort, not a free by-product.
- Data, not just geometry. Asset attributes for a CAFM or FM handover is its own scope. See digital twin and handover data.
Why per-square-foot pricing still exists
Because it is a useful shorthand once the five variables above are fixed. A rate of X per square foot is meaningful for "warehouse, structure and primary MEP, LOD 300, 25mm tolerance, empty, Revit native". Move any of those and the rate stops describing the job.
Treat a per-square-foot number quoted before those are settled as a placeholder. It is not a price; it is an invitation to a conversation about scope.
How to get comparable quotes
If you want three numbers you can actually compare, give every vendor the same six facts:
- Tolerance, per element category.
- An element list — what is in, what is explicitly out.
- Site conditions — occupancy, access hours, expected obstruction.
- Target LOD, per package.
- Deliverable format and the software version it must open in.
- Whether drawings, quantities or asset data come off the model.
Vendors who ask these questions unprompted are pricing the actual job. Vendors who quote from floor area alone are pricing a guess, and the difference surfaces later as a variation.
For what a scoped package involves end to end, see scan to BIM. Where the output feeds a coordination effort, BIM coordination and clash detection is usually the next step.
Scan to BIM cost — common questions.
How much does scan to BIM cost per square foot?
A per-square-foot rate is only meaningful once tolerance, element scope, site occupancy, target LOD and deliverable format are fixed. The same building can differ several-fold across those variables, so a rate quoted before they are settled is a placeholder rather than a price.
What is the biggest cost driver in scan to BIM?
Tolerance, followed by the element list. Modelling to fit-out accuracy and modelling to prefabrication accuracy produce similar-looking geometry from very different amounts of scanning, registration and modelling effort.
Does an occupied building cost more to scan?
Yes, in two ways: restricted or out-of-hours access adds site time, and furniture and equipment block the scanner. Those occlusions leave gaps that have to be resolved by inference rather than measurement, which is slower and less accurate.
Is a point cloud enough, or do I need a model?
A point cloud is a measurement record; it is enough for checking dimensions and clash-testing against a proposal. A model is required when you need scheduled, parametric elements to design, quantify or fabricate against.