Quick summary
- A 3D laser scanning survey is a measured record of what is physically there. It is not a legal boundary survey. In Ontario, only a licensed Ontario Land Surveyor can establish where your property line is.
- What decides whether your data is usable is not the scanner. It is whether the capture was tied to a control network and to a coordinate system the rest of your team already works in.
- Registration tolerance and measurement accuracy are two different numbers. Ask a vendor which one is on the proposal, because they are often quoted as if they were the same thing.
- Decide what you need out the other end, a point cloud, 2D drawings or a Revit model, before the crew is booked. It changes how the building gets captured, not just what gets drawn afterwards.
A scanning survey and a land survey are not the same purchase
When someone asks us for a “laser scanning survey,” they usually mean one of three different things. A measured record of a building’s existing conditions. A topographic surface of a site. Or a legal survey that establishes where the property boundary sits.
The first two are reality capture work. The third is a licensed profession with its own statute. In Ontario, cadastral surveying is defined as “advising on, reporting on, conducting or supervising the conducting of surveys to establish, locate, define or describe lines, boundaries or corners of parcels of land,” and the Association of Ontario Land Surveyors is clear that only its licensed members can carry out that work in the province.
This matters more than it sounds. If your lender, your municipality or your title insurer is asking for a survey, a point cloud will not satisfy them no matter how accurate it is. If your design team is asking for a survey because the drawings are forty years old and nobody trusts the ceiling heights, then reality capture is exactly the right purchase. Getting this wrong at the quoting stage is one of the more common ways a project loses two weeks.
What happens before the scanner turns on
Most of the useful decisions on a scanning job are made before anything gets measured.
A terrestrial scanner sees the building from wherever it is standing. One position gives you one vantage point with shadows behind every column and duct. A usable dataset is dozens or hundreds of those positions stitched into a single coordinate space, and the stitching is the part that has to be controlled rather than assumed.
That is what the numbered targets are for. Our crews place them so adjacent scan positions share enough common geometry to lock together, and so the network carries through the building rather than drifting down a long corridor. In the photo below, targets A-03 through A-05 sit along temporary hoarding at roughly the height the scanner will see them from both sides of the opening. They are not decoration and they are not placed at random.
If those targets are also tied to points with known coordinates, the whole dataset inherits that coordinate system. If they are not, you get an internally consistent model of the building that is floating in space with no relationship to anything outside it. Both are legitimate outcomes. Only one of them is useful if the steel fabricator needs to work from your gridlines.
Site grid or geodetic coordinates: decide before the crew arrives
There are three practical answers to “what should this line up with,” and the right one depends on who uses the data after us.
A local system
The building defines its own origin. Fine for a standalone record of existing conditions, a heritage archive, or a fit-out where nothing outside the demising walls matters.
The project or site grid
The gridlines the design team and the general contractor are already using. This is what you want for construction verification, for anything that will be compared against a design model, and for fabrication. It is also the one people forget to specify, because on site the grid feels so obvious that nobody thinks to write it into a scope.
Geodetic coordinates
A published system such as a UTM or MTM zone on NAD83. Needed when the work ties into municipal data, civil design, infrastructure, or anything that has to sit correctly relative to the world rather than relative to the building.
The cost of deciding late is that the data was never tied to anything in the first place. Transforming a cloud after the fact is possible when you have enough common points to solve from, and painful when you do not. On a live site it usually means sending a crew back.
The same question comes back later inside Revit, where a model carries a project base point and a survey point that do the same job at the modelling end. If your consultants are sharing coordinates between models, the scan has to arrive already speaking that language.
What you actually receive
A scanning survey is not one deliverable. It is a capture, and then whatever gets built from that capture. Deciding which of these you need before the site visit changes how densely we scan, where we set up, and how much time we spend on areas that would otherwise be background.
| Deliverable | Format | What it is for | Typical turnaround |
|---|---|---|---|
| Registered point cloud | .E57, .RCP / .RCS | The measured source of truth. Everything else is derived from it. | Within 48 hours |
| 2D as-built drawings | DWG, PDF | Floor plans, elevations, sections, BOMA area calculations | 1 to 3 weeks |
| Scan-to-BIM model | RVT, LOD 200 to 400 | Design coordination, clash detection, renovation and fit-out | 1 to 3 weeks |
| Deviation analysis | Report and colour map | Comparing what was built against what was designed | Scope dependent |
| 360 virtual tour | Web viewer | Walking the site remotely without measuring from it | Scope dependent |
Turnaround on modelling moves with level of development and floor area, which is why the range above is a range. Our crews are typically on site within one to three days of booking, and the registered cloud follows within 48 hours of leaving.
If you are weighing drawings against a model, we have written a longer comparison of Scan-to-CAD and Scan-to-BIM as deliverables, and a separate guide to choosing a level of development.
Registration tolerance and measurement accuracy are two different numbers
This is the part worth reading twice, because it is where proposals get compared badly.
The scanner specification
A scanner’s published specification describes how precisely that instrument measures a single range under test conditions. It is a property of the hardware and it is the number most often quoted in marketing.
Registration tolerance
Registration tolerance is different. It describes how well the individual scan positions agreed with each other when they were stitched together on your building, and it comes out of the registration report for your job. Two crews using the same scanner can produce very different registration results depending on how the control network was laid out.
Modelled accuracy
Then there is the accuracy of a modelled element, which is different again. A wall in a Revit model is a designer’s decision about where a slightly irregular real surface should be represented as a plane. The point cloud can be tight and the model can still simplify.
The U.S. Institute of Building Documentation’s Level of Accuracy specification exists precisely to stop these being conflated, and its current version introduces standard deviation as the basis for expressing tolerance rather than a single headline figure. As of September 2026 it is the clearest common language available for writing accuracy into a scope of work.
So when you are handed a number, ask which one it is. If the answer is a specification sheet rather than a figure from the registration report for your building, you do not yet know what you are buying. We cover the wider question of accuracy standards and quality control in more detail separately.
Where you still need a licensed surveyor
Reality capture does not replace cadastral work, and any vendor who implies otherwise is selling you a problem.
You need a licensed surveyor for property boundaries and corners, for legal descriptions, for anything that will be registered on title, and for plans that require a surveyor’s seal. In Ontario that means an Ontario Land Surveyor. Other provinces and states have their own equivalent.
The arrangement that works well is the ordinary one: the surveyor establishes and certifies control, and we tie the scan to it. You end up with a measured record of the building that sits correctly in a coordinate system somebody is professionally accountable for. On projects where a surveyor is already engaged, ask them for the control before scoping the scan rather than after.
My perspective: two jobs that taught us to ask first
Brooklyn: the coordinate system that never arrived
The first was a restoration project in Brooklyn. We had scanned the site and were working in Civil 3D, and both the point cloud and the model needed to be tied to a survey coordinate point the municipality was going to provide. The surveyor on that project never gave us the coordinate system. The city’s submission required the model to be tied to it, and so did our contract, so we were holding finished work we could not hand over.
What saved it was that the targets were still up. We had left them in place after capture, so when we brought in a second surveyor they were able to pick up those same target positions and give us coordinates against them. That let us merge the scan data with the surveyed targets and georeference the whole thing after the fact. If those targets had come down when our crew left site, the only remaining option would have been to scan the building again.
Old Toronto: a flatness analysis with no origin
The second was a floor flatness analysis on a pickleball facility in old Toronto. The client had no datum, and no survey coordinates for one either. So when we delivered, there was no origin point we could attach the results to. The client and the construction crew were left holding a flatness analysis with nowhere to put it, and it caused a real upset. None of it was a measurement problem.
What we do now
Both jobs came down to the same gap. The scan, the use the data was headed for, and the survey system it had to live in were being treated as three separate conversations, and nobody had all three at the same time.
So here is what we do now. We ask for survey coordinates before the job is booked. If a client wants scan data merged with survey data, they need to tell us that up front and provide the dataset. And where the answer genuinely is not available yet, we set our own control points while we are on site and we leave them in place, so the surveyor can pick them up afterwards and send the coordinates back to us. Then we merge the cloud, the control and the model, and the whole package is georeferenced.
One practical thing worth doing whenever you can: put the scanning targets and the surveyor’s targets on the same points, so both sets of data share an origin. It costs nothing on the day and it removes this entire category of problem.
Questions to ask before you sign the scope
- What coordinate system will the deliverable be in, and who is providing the control?
- Is the accuracy figure on this proposal a scanner specification or a registration result?
- Will I receive the registration report, or only the finished cloud?
- Which areas are excluded because they cannot be seen or accessed, and how will that be recorded?
- What level of development is the model, and which systems are modelled at that level?
- If a licensed survey is also required, who is engaging the surveyor?
A vendor who answers all six quickly has done this before. A vendor who treats the first one as a detail to sort out later is telling you something useful.
Frequently asked questions
Is a 3D laser scanning survey the same as a land survey?
No. A laser scanning survey is a measured record of physical conditions, delivered as a point cloud and whatever drawings or models are built from it. A land survey, in the legal sense of establishing boundaries, is restricted work. In Ontario only a licensed Ontario Land Surveyor can carry out cadastral surveys. The two are frequently used together, with the surveyor providing control that the scan is tied to.
Do I need to supply survey control before the scan?
Only if the deliverable has to sit in a specific coordinate system. If the point cloud needs to line up with a project grid, a civil model, or municipal data, then control has to exist before capture and the targets get tied to it. If the building is being recorded for its own sake, a local system is enough. The decision belongs at scoping, not after delivery.
What accuracy should I ask for?
Work backwards from the decision the data supports. Fabrication and clash detection need tighter tolerances than a space plan or a facility record. Rather than asking for the tightest number available, state what the deliverable will be used for and let the tolerance follow from that. The USIBD Level of Accuracy specification gives you standard language for writing it into a scope.
How long does a scanning survey take?
Time on site depends on floor area, how cluttered the space is and how much of it is occupied while we work. The registered point cloud follows within 48 hours of leaving site. Drawings and Revit models take one to three weeks depending on level of development and size.
Can you scan a building that is occupied and operating?
Yes, and most of what we scan is occupied. It changes how the capture is planned rather than whether it is possible. Occupied space means more scan positions to work around obstructions, and it means agreeing in advance which areas will be accessible and which will be recorded as excluded.
Getting the scope right
The difference between a scan you build from and a scan you look at is decided before anyone opens a tripod. Coordinate system, control, accuracy basis and final deliverable. Four questions, about ten minutes, and they decide whether the data is worth what you paid for it.
Answer all four before you go out for pricing and you get proposals you can actually compare, instead of four numbers measuring four different things.
Our 3D laser scanning services page sets out the full range of capture and deliverables, and the as-built documentation and Scan-to-BIM pages go deeper on the two most common outputs.
