The short answer
When a PDF drawing and an IFC model describe the same physical items, their quantities are alternative sources—not extra items to add together. In Soban Labs, give matching contributions the same item name and unit, keep their source references, then choose the final source in the takeoff register.
Combine them only after confirming that they cover separate scopes. A final quantity still needs your review of the source, units, revision and assumptions.
1. Agree the item and scope first
Define what you are counting or measuring: for example, “W2 windows, Level 1, current revision.” A matching name alone does not establish matching scope. Check the level, zone, type and inclusions in both files.
Choose an appropriate quantity basis. Windows may be counted in ea; a beam may need length in m, along with its material and section/profile to make the item meaningful. Concrete may require volume instead. These are starting points, not universal estimating rules.
Soban groups contributions by item name and unit. Counts, lengths, areas and volumes remain separate. Use distinct names for separate specifications or scopes; do not group different beam profiles under one generic “Beam” item.
2. Check the drawing contribution
Open the PDF in Drawing Lab. For lengths and areas, calibrate each relevant sheet from a known dimension and verify the units. Counts do not depend on drawing scale, but still depend on identifying the right symbols and scope.
Use one consistent item name, such as W2, for the matching marks. Review repeated counts, excluded items and sheet coverage. If you use Suggest, inspect its candidates before accepting them; a suggestion is not a checked quantity.
Choose Review totals to inspect grouped quantities. Then open Export & share → Add to takeoff register. Check the destination, groups, names and assumptions before saving. Unscaled measurements and marks still needing revision review are excluded from the drawing contribution.
3. Check the model contribution
Open the matching IFC in Model Lab and use Data → Quantities. Set the intended element scope and category. Check the element type, material and section/profile or other required specification, and select the measurement basis you intend to use.
An element-record count tells you how many selected records exist; it does not establish that they are the right windows, beams or complete assemblies. Reported lengths, areas, volumes and masses depend on the quantities and units supplied in the IFC. Soban does not infer missing takeoff values from geometry.
Review and correct information where needed, mark the intended records reviewed, then use Add reviewed to register to add them to the register. Local corrections do not rewrite the IFC or automatically update an earlier contribution. Section cuts do not reduce full-element quantities. GLB viewing does not provide IFC takeoff data.
4. Choose the final source
Illustrative worked example
| Saved source | Count |
|---|---|
| Drawing contributionFive checked W2 marks on the relevant sheet | 5 ea |
| Model contributionFive checked W2 window records in the matching scope | 5 ea |
| Final quantityReviewer chooses the model source | 5 ea |
The result is 5 ea, not 10 ea. Both contributions remain as evidence; they refer to the same windows.
These numbers are invented to explain reconciliation, not measured results from a client project or a benchmark. The built-in drawing and model samples are separate examples, not a matched building dataset.In the takeoff register, open the item and inspect Source references and assumptions. For this example, choose Use model quantities as the final quantity source and save the decision. Choosing the drawing source would also give 5 ea here. Matching totals alone do not prove that either source is complete.
If the drawing says five and the model says four, pause to check scope, revisions, duplicates and missing elements. Do not average the counts. If you deliberately select one source, record why. Combine only genuinely separate scopes—for example, two explicitly separate zones—not two representations of the same zone.
5. Keep gaps visible
A blank or unresolved final quantity is not zero. The register flags and excludes contributions from missing or older shared file revisions, and flags conflicting repeated references. Replace or remove the problematic contribution after review; the latest model is not silently substituted.
When a model lacks a usable length or volume, use a suitable checked drawing contribution or obtain corrected model data. A successful file load, a green review colour or a completed mission does not certify a quantity. Changes to source files or local corrections need a deliberate new review and contribution.
6. Export the evidence with the result
Review the final quantity, unit, source policy and assumptions, then download Quantity CSV from the register. Each item has one Item total row, followed by its Source reference rows. Those reference rows are evidence behind the total; do not sum them again as additional final quantities. Unresolved final quantities stay blank.
Before relying on the schedule
- Correct file revision, item specification and physical scope?
- Appropriate quantity basis and units?
- Missing information, excluded records and discrepancies reviewed?
- Final source chosen without counting the same items twice?
- Source references and assumptions retained?
Try one item before a whole project
Use matching PDF and IFC files you are authorized to process, or practice the two labs separately with their public samples. Check one clearly defined item, trace its sources and export its register entry before expanding the scope.
Already using a connected assistant? Ask it to find saved quantity groups and show their source references. The current MCP briefing and search tools read saved Team records; they do not inspect the contents of drawings or models. See the AI connection guide.
This guide describes the Soban Labs beta workflow available on October 19, 2026. It explains quantity review and source reconciliation; it is not engineering certification or a substitute for project-specific measurement rules.