A bill of quantities turns a drawing set into measured items that can be priced. Done well it produces comparable tenders and a sound cost picture. Done carelessly it produces disputes, because the arguments are almost always about what was assumed rather than what was measured.
What a take-off involves
The measurer works systematically through the drawings, extracting quantities by element: areas of wall by type, volumes of concrete, lengths of skirting, numbers of doors, and so on. Each item carries a description precise enough that a contractor prices the same thing you measured.
The description matters as much as the number. "Partition, 100mm metal stud, 12.5mm board both sides, taped and filled, 3m high" is priceable. "Partitions, 400 square metres" is not.
Where accuracy actually comes from
Drawing quality. You cannot measure what is not drawn. Incomplete sets produce assumption heavy take-offs.
Consistent measurement rules. Whether openings are deducted, how junctions are treated, whether finishes are measured to structure or to finish. A standard method keeps this consistent, and where none is adopted the rules should be stated explicitly.
Documented assumptions. Every take-off contains judgement calls where drawings are silent. These belong in a written list beside the bill, not in the measurer's head. Most disputes are traceable to an assumption nobody recorded.
What a bill is used for
Comparing tenders on a like for like basis, so differences reflect price rather than interpretation. Establishing a cost plan early. Valuing variations against agreed rates. Assessing progress payments against measured work.
Common problems
- Measuring from an out of date drawing revision, which invalidates everything downstream
- Double counting where elements appear on multiple drawings
- Omitting items shown only in specification and not on drawings
- Descriptions too vague to price consistently
- Unrecorded assumptions that surface as claims later
Models help, but not automatically
Quantities can be extracted from a BIM model, which is faster and reduces arithmetic error. But the output is only as reliable as the modelling discipline behind it. A model where similar elements are inconsistently classified produces confident nonsense. Model derived quantities need checking against drawings, not trusted blindly.
What to provide for a good take-off
A complete drawing set at a known revision, the specification, any schedules, and clarity on the measurement rules to be used. Where information is missing, ask to be told what was assumed rather than leaving the measurer to guess silently. The assumptions list is the most useful part of the deliverable.