Every scan-to-BIM deliverable looks great. The delivered dataset loads instantly, the model loads into place, and the measurement lands right on the edge of the beam. Then your as-built data goes in: 400 scans from three survey campaigns, a federated mesh and that’s when you find out what you actually bought.
This guide gives contractors, design/build teams and owners a practical way to evaluate scan-to-BIM software and services, built around the two questions that decide most use cases: will it be trustworthy, and will my model meet requirements?
Those are the right questions, and no vendor can answer them for you in a slide deck. Accuracy depends on your as-built data, and more importantly the technicians interpretation. Alignment depends on your coordinate systems, your registration quality and how your models were set up. The only honest answer comes from testing the results on the laser scanner data, your own scans and your own interpreatation
It helps to be specific about what each question really means on a project.
The level of detail required. It's how long it takes to get registered scans published, how quickly a coordinator can open and move through a large site, how responsive Revit stays when the point cloud is linked, and how many days pass before a trade partner sees the data. A platform can be quick at one of these and slow at the rest.
"Lines up" isn't a yes or no. It means the scans and the model share the same origin and orientation, measurements hold within the tolerance your project agreed, and that alignment survives when the data moves back into Revit or Navisworks. If any one of those breaks, your team ends up checking everything twice.
The rest of this guide turns both questions into tests you can run in a two-to-four-week pilot.
A scan-to-BIM service delivers a finished model built by a provider’s modelers. Scan-to-BIM software lets your own team register scans and build the model. A visualization and control tool doesn’t create models: it shows the registered scans next to the delivered model so you can check one against the other.
| Scan-to-BIM service | Modeling and registration software | Visualization and control tool |
|
| What you get | A finished as-built model, delivered by a provider | Desktop tools to register scans and build the model in-house | A shared view of the registered scans next to the delivered model |
| Who builds the model | The provider's modelers | Your in-house team | Nobody. It doesn't author models |
| What it's good at |
Capacity when you don't have modelers | Full control if you have the skills in-house | Checking that the model matches reality, and sharing both with the project |
| What it doesn't do |
Show you how the scans were interpreted | Make the data easy to share outside the BIM team | Register scans or create geometry |
| Key question to ask |
How will I check what I receive? | Who on my team will use it, and how often? | Can it show me where the model and the scan disagree? |
Most contractors and owners outsource the modeling at least some of the time. That's where the two questions in this article really come from. Whether the delivered model is faithful to the site is much harder to see, and it's the one that costs money later.
A scan-to-BIM model is never a direct copy of the point cloud. Someone sits in front of millions of points and decides what they are. Is that a 150 mm pipe or a 200 mm one? Is the slab level or does it fall toward the drain? Is that grey shape a valve worth modeling, or noise worth ignoring?
Every one of those calls is an interpretation. When you outsource the work, you don't choose who makes them, how experienced they are, or how motivated they are to chase the details on a fixed-price job. Two modelers can work from the same scan and hand you two different models. Both will look clean in Revit.
That's why "will my model line up?" is the right question. A model that looks clean but simplifies a congested ceiling void, misses a beam or places a riser 60 mm off just moves the problem to site, where it shows up as a clash, a prefab spool that doesn't fit, or a change order.
The usual control is a visual spot check and a lot of trust. The better control is to keep the registered scan next to the delivered model, so you can check the modeler's interpretation after delivery, before anyone designs or fabricates against it. Our article on scan-to-BIM services covers what to expect from the provider side.
You can't check a model against a vague brief. If the contract says "as-built model of Level 2," every interpretation is technically correct. Before you choose a provider or a tool, write down three things.
| Spec | What it defines | Example in a brief |
| Level of detail (LOD) | How much geometry gets modeled, and how precisely it's shaped | MEP mains and branches down to 50 mm, with valves and dampers as families |
| Level of information (LOI) | Which data each element carries | System, size, material, and tag or asset ID for equipment |
| Accuracy and tolerance | How close the model has to be to the scan | System, size, material, and tag or asset ID for equipment |
LOD and LOI work together. A duct modeled at the right size but with no system data fails the LOI. A pump with perfect data but placed 200 mm off fails on accuracy. Both get through a quick visual review.
Two practical tips. First, set different tolerances by discipline: what's fine for a partition wall isn't fine for a connection a prefab module will bolt to. Second, list the areas where interpretation is hardest, such as congested ceiling voids, plant rooms and risers, and tell the provider those will be checked first.
Alignment matters too. The scans and the model need to share the same coordinate system before any comparison means anything, so agree on survey control or a shared reference frame with your survey team up front. Cintoo's scan and model alignment guidance walks through the options.
If you're evaluating software to control model quality, these are the criteria that matter. What you need is to see the site clearly and compare it honestly with what was delivered.
The best test is still a pilot. Take a model you've already received, ideally one that caused a problem on site, load it next to its scans and see how quickly the issue becomes visible.
Once the model arrives, the goal is to answer one question: does it faithfully represent what was scanned, at the LOD and LOI you asked for? Cintoo is the QA/QC layer between the scan and the model. It doesn't author models or register scans; it lets you compare the delivered BIM or CAD model with the registered scans, measure, and document what doesn't match. Among Cintoo clients, the review runs as a loop, whether the model comes from an in-house team, a design office or a subcontractor:
The QA/QC loop in Cintoo: each finding is annotated, assigned, fixed and rechecked on the new model version until the model is signed off.
This is also where you learn how good your provider really is. After two or three projects, you'll know which ones chase the details and which ones don't.
Throughout the loop, reviewers check against the actual site, not a thinned-out version: Cintoo preserves high-resolution reality capture and streams it in a browser, and the provider, designers, site team and owner all review the same data with role-based access. Cintoo is certified to ISO/IEC 27001:2022 and SOC 2 Type 2, and complies with GDPR. On site, the same platform also supports 3D scan progress monitoring of BIM elements against new scans.
A word on expectations: Cintoo measures how far the model sits from the scan, but it doesn't sign the model off for you. Results depend on the quality and registration of the source scans and the model. What it gives you is the evidence to judge for yourself. The full list is on the Cintoo feature list, and for a project example, read how Hensel Phelps planned, built and managed a large construction project with Cintoo.
It depends on how often you scan and whether you have modelers. Outsourcing gives you capacity. Modeling in-house gives you control. If you outsource, plan how you'll check the delivered model against the scans before you need it.
LOD (level of detail) defines how much geometry is modeled and how precisely. LOI (level of information) defines what data each element carries, such as system, size or asset tag. A good brief specifies both, plus a tolerance.
Usually it's one of three things: the model and scans aren't in the same coordinate system, the brief left too much room for interpretation, or the modeler simplified hard-to-read areas. Check alignment first, then review the congested zones.
Partly. Geometric deviation, yes: Cintoo's comparison tool shows where model elements depart from the scan. Whether the model meets the LOD and LOI in the brief still needs an experienced reviewer, and with Cintoo that reviewer can control the delivered model against the scan before it goes into production.
No. Cintoo doesn't create models or register scans. It's a cloud platform to visualize and compare delivered models against registered scan data, and share the fi