QTO as the single source of truth for variations

A variation lands and the ritual begins. The contractor's QS measures one number. The cost consultant measures another. Three weeks of correspondence follow about whose number is right, and none of it is about the work, which everyone agrees needs doing. The dispute is not about the change. It is about the measurement.

Both sides measured honestly. They just measured different artefacts at different times, and that fork, not bad faith, is where variation disputes come from.

Two sources, one argument

The fork happens early. A BOQ gets measured from tender drawings. The design moves on. The site remeasures what it actually installs. By mid-project there are three versions of every quantity, drawing-derived, model-derived, site-derived, and each lives in someone's spreadsheet, maintained in good faith, drifting from the others with every revision. When a variation needs pricing, each party reaches for its own version, and the gap between the versions becomes the negotiating range.

Re-measuring is also where the cost goes. One wall-thickness change, one service-zone shift, and a manual take-off starts again from zero, so by the time the BOQ is current, the model has moved again. The lag is structural. No amount of measuring faster fixes a workflow where the measurement and the design live in different places.

Make the model the measurement

The fix is to collapse the fork: quantities extracted element-by-element from the coordinated model, mapped to your cost codes, versioned in the CDE alongside the geometry they came from. Not a take-off of the design, a property of the design. ISO 19650 already thinks this way: the level of information need treats alphanumerical data, including quantities, as part of the element's information, not a separate document someone maintains by hand.

Then a design change stops being a re-measuring event. Extract the revised model and the variation is the difference between two versions: which elements were added, which removed, which modified, each carrying its quantities and cost codes. The commercial question stops being "whose number is right", a question with no answer, and becomes "which revision are we pricing", a question with one.

What it takes to hold

Single-source quantities are a discipline before they are a tool. Three things have to hold. Elements carry consistent classification and cost codes, a system like Uniclass or OmniClass embedded in the family alongside the cost code, because an unclassified element is a quantity that escapes the net. Extraction runs at every milestone revision, not on demand, a delta only means something between two named versions. And the trail stays auditable: any line in the BOQ traces back to the elements that produced it, so a challenged number is answered by opening the model, not by re-measuring.

None of this needs exotic software, extraction workflows in tools like CostX, Assemble, or Revit schedules exported for the quantities do the mechanical part. What it needs is the model treated as the commercial record, not just the design record, from tender onwards.

The quiet benefit

Projects that price variations as deltas argue less, and not because anyone became more reasonable, because the thing being argued about stopped existing. The numbers come from the source both parties already accepted when they accepted the model. That is what model-based QTO is actually for: not faster take-offs, although you get those, but a commercial conversation with one set of numbers in it.