Construction projects are increasingly documented in three dimensions. Not as a novelty, and not as a marketing exercise — as a working record of what was built, when, and to what tolerance. The weekly drone flight has joined the daily diary as a routine instrument of site management.
The technology side of this is largely solved. Drones fly, LiDAR scans, point clouds process, models render. The harder problem is the one most projects still get wrong: getting the resulting data into the hands of the people who need it, in a form they can actually use.
Here’s how 3D mapping fits into construction work in 2026, what teams actually produce, and the bottleneck that quietly throttles most of the value.
What 3D mapping does on a construction site
Three uses dominate, and they map cleanly to different phases of a project.
Progress tracking
Weekly or fortnightly drone flights produce orthomosaics and elevation models that document earthworks, formwork, structural progress, and site layout. Two captures from different dates can be overlaid to measure cut and fill volumes, monitor the rate of progress, and produce defensible visual evidence for monthly progress claims.
The value is not the individual capture. The value is the time series. A single orthomosaic is a picture. Twelve consecutive monthly orthomosaics are a record of the project from clearing through to handover, and that record is referenceable for the life of the asset.
As-built verification
Terrestrial LiDAR scans of completed structural elements — walls, floors, MEP risers, façades — capture the geometry of what was actually built, which is rarely identical to what was designed. The as-built point cloud can be overlaid onto the design model to flag deviations before they propagate downstream into finishes, fitout, or systems coordination.
This matters most on projects where tolerances are tight: pharmaceutical clean rooms, data centres, fabrication facilities, anything with prefabricated finishes that have to fit existing site conditions exactly.
BIM coordination and clash detection
A point cloud of existing site conditions overlaid with the federated design model surfaces clashes during design review rather than during construction. An existing service that conflicts with a proposed footing, a finished floor level that does not match the as-built survey, a structural opening that landed two metres from where the model says it should be — all of these are easier to fix on a Tuesday in design review than on a Wednesday in the ground.
For a primer on the underlying data types, see what 3D mapping is and what a point cloud is.
The capture technologies in use
Different parts of a construction site demand different tools. Most projects of any size end up using two or three in combination.
| Technology | Best for | Typical accuracy | Capture rate |
|---|---|---|---|
| Drone photogrammetry | Earthworks, site overview, façades | 20–50 mm | 10–50 ha per flight |
| Drone LiDAR | Vegetated sites, bare-earth DTMs | 30–80 mm | 5–30 ha per flight |
| Terrestrial LiDAR | Interiors, structural details, MEP | 2–6 mm | 30–80 scans per day |
| Mobile mapping | Corridors, tunnels, large interiors | 20–50 mm | 1–5 km per shift |
| 360° photography | Visual record, progress diary | Visual only | 100–300 stations per day |
Drone photogrammetry is the workhorse for most general contractors and surveyors. It is cheap to deploy, fast to capture, and the outputs (orthomosaic, DSM, point cloud, video) cover the majority of progress-tracking needs.
Terrestrial LiDAR enters the picture when millimetre accuracy matters — typically inside structures, around MEP, or at façade detail. Mobile mapping covers everything in between, and is increasingly used for long internal corridors, multi-level car parks, and tunnel projects where setting up tripod stations would take days.
360° photography is the most underrated tool of the four. It captures no geometry but produces a comprehensive visual record that anyone on the project can navigate without training. For monthly progress documentation, a Matterport or 360° camera walk-through is often more useful to the project team than a point cloud.
Typical deliverables
A modern construction survey delivery includes some combination of:
| Deliverable | Format | Primary user |
|---|---|---|
| Orthomosaic | GeoTIFF | Project manager, client, planning |
| Digital terrain model | GeoTIFF | Civil engineer, earthworks contractor |
| Point cloud | LAS, LAZ, E57 | Surveyor, structural engineer, BIM coordinator |
| Mesh / 3D Tiles | OBJ, GLB, 3D Tiles | Client presentations, design review |
| As-built CAD | DXF, DWG | Designers, services coordinators |
| IFC model | IFC | BIM coordinators, asset owners |
| Drone video | MP4 | Marketing, client updates, dispute evidence |
| 360° virtual tour | Equirectangular JPG | Site walkthrough, remote inspections |
| Survey report | Everyone |
The list has grown longer over the last decade. Five years ago, the deliverable was a DWG and a PDF. Today, the DWG and PDF are the two files anyone in the project team can reliably open — and the rest sit unopened in a shared folder unless someone provides a viewer.
See what file formats drone surveys produce for a deeper breakdown.
The stakeholder problem
The fundamental challenge with construction spatial data is not the data. It is the audience. A single capture is consumed by at least six distinct roles, each of whom needs a different view:
- Site supervisor wants the orthomosaic on a tablet on site, to compare against what is in front of them.
- Project engineer wants the point cloud overlaid on the design model in BIM software.
- Project manager wants the visual record and a comparison with the previous month’s capture.
- Client representative wants a high-level walkthrough they can show to their executive committee.
- Insurer wants documentary evidence of conditions on a specific date, accessible without specialist tools.
- Regulator wants the underlying data, not a screenshot of it.
A single Dropbox link cannot serve all six. The site supervisor cannot open the LAS file on the tablet. The client representative does not want to install CloudCompare. The insurer’s claims adjuster has never heard of QGIS.
The result, on most projects, is that the spatial data ends up being seen by one or two technical people and converted into screenshots and PDF excerpts for everyone else. The expensive part of the deliverable is consumed by the cheap workflow of cropping images into a Word document.
For a deeper look at where this breaks down, see file sharing for construction.
Capture cadence
Most construction projects settle into one of three rhythms.
Routine cadence — typically weekly or fortnightly drone flights over the active works area, producing orthomosaics and point clouds for progress tracking and volumes. This is the dominant pattern on infrastructure projects, large residential, and any earthworks-heavy job.
Milestone cadence — captures triggered by specific construction events: pre-pour scans of formwork, post-demolition surveys, end-of-stage as-builts, defects period documentation. Each capture is a snapshot tied to a contractual or technical event.
Monitoring cadence — continuous or near-continuous capture for specific risks: slope stability on excavation works, settlement monitoring around heritage neighbours, vibration-sensitive structures. Lower spatial coverage but much higher temporal density.
Most projects run two of these in parallel. A typical infrastructure project will have weekly drone flights (routine), milestone scans at the end of each major stage (milestone), and dedicated monitoring of any specific risk (settlement on a heritage retaining wall, for instance).
The implication for data management is that a single project will accumulate dozens of capture sessions over its life, each producing several gigabytes of deliverables. A two-year project doing weekly flights generates over a hundred captures. By handover, the project has produced more spatial data than the design drawings, by an order of magnitude.
File volumes in practice
Order-of-magnitude figures for a typical mid-sized construction project:
| Capture type | Files per session | Volume per session |
|---|---|---|
| Drone photogrammetry (10 ha site) | 4–6 outputs + raw images | 8–25 GB |
| Drone LiDAR (10 ha site) | 2–4 outputs + raw scans | 15–60 GB |
| Terrestrial LiDAR (structural detail) | 40–100 scans + registered cloud | 20–80 GB |
| Mobile mapping (1 km corridor) | 1–2 clouds + raw | 30–100 GB |
| 360° walkthrough (10,000 m²) | 300–500 panoramas | 5–15 GB |
A two-year project with weekly drone flights, monthly terrestrial scans, and quarterly mobile mapping will accumulate 1–3 TB of spatial data by handover. This is not a niche edge case. This is the routine output of a modern construction project, and most projects have nowhere coherent to put it.
Storage and sharing over a multi-year project
Construction projects last years. Stakeholders rotate through. The site engineer who commissioned the original survey is on a different project by the time the as-built scan is being referenced for a defects claim. The client representative who attended the design review has moved roles. The contractor’s BIM coordinator has left the firm.
Generic file storage does not handle this well. Folders get renamed, restructured, and migrated between platforms. Files get archived to cold storage. Permissions get revoked when a contractor’s licence lapses. Three years into a project, “where is the April 2024 ortho” is a half-day’s expedition into nested SharePoint sites that nobody remembers building.
The data deserves better treatment than this. Each capture should be anchored to a specific site, time-stamped to a specific date, and accessible to anyone with a current need to view it — without requiring them to install software, hunt through folders, or chase the original surveyor for screenshots.
This is the gap Swyvl is built to close. A construction site gets its own site record. Every capture lands on that site, tagged to a date and viewable in the browser. The site supervisor opens the orthomosaic on a tablet. The BIM coordinator opens the point cloud in a desktop browser overlaid with their design model. The client representative gets a branded share link. Same data, six different views, no software installs.
Where to start
If your project is currently delivering spatial data via WeTransfer and Dropbox links, the upgrade path is straightforward.
Keep the deliverables that already work in your existing channels — DWG files, PDF reports, project photos — moving through whatever platform your project uses for documents. Procore, Aconex, SharePoint, and Autodesk Construction Cloud handle these fine.
Move the spatial deliverables — orthomosaics, point clouds, 3D models, drone video, 360° photos — to a platform that can actually display them. The first time a site supervisor opens a point cloud on their tablet at the job face, the change pays for itself.
For more on how this transition looks in practice, see how to deliver drone survey data and best survey data delivery platforms.
3D mapping is no longer the speciality offering it was a decade ago. It is becoming the default way that construction projects of any significant size are documented, monitored, and verified. The capture technology is mature. The processing software is mature. What is still maturing — and what determines whether the data delivers its value — is the delivery layer. That is the part still worth investing in.