Drawing QA/QC is the review a construction drawing set goes through before issue, checking that each sheet is internally correct, that sheets agree with each other, and that the set conforms to the applicable standards and codes. Most errors that reach site are not drafting mistakes on a single sheet. They are disagreements between sheets that nobody had time to cross-check.
This is a working checklist, organised by what is actually being compared against what.
What drawing QA/QC actually means
Two terms get used interchangeably and should not be.
Quality assurance is the process: the standards, templates, naming conventions and review stages that make a good set likely. It happens before anyone draws anything.
Quality control is the inspection: checking the set you actually produced against those standards before it goes out. It happens at the end, under time pressure, which is why it is the half that gets cut.
A drawing review is quality control. It answers one question: if we issue this set today, what will come back?
Why the review pays for itself
The cost of a missed check is not the correction. It is the question it generates and the work already in place when the answer arrives.
| Finding | Source |
|---|---|
| The average cost to review and respond to a single RFI was $1,080, with a median response time of 9.7 days and an average of 796 RFIs per project | Navigant Construction Forum, 2013 study of roughly one million RFIs across 1,300 projects |
| Poor project data and miscommunication cause 48% of all rework in the US, around $31.3 billion a year | FMI and Autodesk, Harnessing the Data Advantage in Construction, 2021, surveying more than 3,900 professionals |
| Direct rework runs about 5% of contract value, closer to 9% once field supervision and overhead are counted | Navigant Construction Forum |
| Large projects typically run 20% longer than scheduled and up to 80% over budget | McKinsey Global Institute |
At those averages, a project's RFI processing alone is roughly $860,000 before a single answer changes anything on site. Not all of that is preventable by drawing review. A meaningful share of it is: an RFI raised because a detail callout pointed at a sheet that was never drawn is a pure self-inflicted cost.
The Navigant RFI figures are over a decade old, so treat them as the documented scale of the problem rather than today's exact price.
Part 1: Title block and sheet setup
The cheapest checks, and the ones that make a set filable and referenceable.
- Every sheet carries a unique sheet number
- Every sheet carries a sheet title matching the drawing register
- The revision and revision date are current
- The scale is stated, or the sheet is marked NTS
- The discipline prefix matches the discipline (A for architectural, S for structural, M/E/P, C for civil)
- North arrow present on every plan sheet
- Graphic scale bar present where the set may be reduced or reprinted
- Consultant and project details are current, not carried over from the last job
Part 2: Annotation, dimensions and tags
Checks that a sheet is internally complete.
- No dimension without a value
- No tag without content, including door, window, room and equipment tags
- Every grid line is labelled
- View titles match what the view actually shows
- Legends and keynotes cover every symbol used on the sheet
- Line weights distinguish primary elements, secondary elements and annotation
- No overlapping or clipped text at issue scale
- Revision clouds match the revision schedule
Part 3: Cross-sheet consistency
This is where reviews fail, and it is the part worth reading twice.
- Every callout points to a sheet that exists in the set
- Every detail reference resolves to the detail number it claims
- No duplicate sheet numbers
- Section and elevation markers point at views that were actually produced
- Room names and numbers agree between plans, schedules and finish schedules
- Door and window marks agree between plans and their schedules
- Levels and datums agree across plans, sections and elevations
- The sheet index matches the sheets actually issued
A callout pointing at a detail nobody drew is invisible on the sheet it sits on. It looks completely correct in isolation. It is only wrong in company, which means catching it requires holding two sheets in your head at once, several hundred times.
Part 4: Cross-discipline coordination
- Grid labels agree between disciplines. Architectural grid C should not be structural grid D
- Levels agree between architectural, structural and MEP
- Ceiling heights agree between architectural sections and MEP coordination drawings
- Structural openings match architectural openings in size and position
- Slab edges and levels agree between architectural and structural
- Discipline drawings share the same north orientation and setting out
Part 5: Drawings against the model
If the set was published from a model, the set and the model drift apart from the moment the first sheet is issued.
- Sheets in the set still match their views in the model
- Title block data agrees with the model's sheet parameters
- Grids in the model agree with grids on the sheet
- Views published are the current view versions, not superseded ones
Part 6: Standards and code conformance
- The set conforms to the firm's drawing standard
- The set conforms to the building code for the jurisdiction
- Client-specific or framework standards are met
- Layer, naming and file conventions match the BIM execution plan
Which of these actually fail
Parts 1 and 2 are tedious but tractable. A reviewer with a fresh eye catches most of them, and they are the checks people think of when they think of drawing QA.
Parts 3 and 4 are where sets fail, for a structural reason rather than a competence one. The number of pairwise comparisons grows with the square of the sheet count. A 40-sheet set has 780 possible sheet pairs. A 400-sheet set has just under 80,000. No reviewer does that arithmetic, so in practice cross-sheet checking gets sampled: the sheets someone remembers being changed, checked against the sheets someone remembers them relating to.
Sampling is a reasonable response to an impossible task. It is also why the errors that survive to site are disproportionately cross-sheet errors, and why they surface as RFIs rather than as redlines.
The checks that are easy to describe are the ones humans do well. The checks that are hard to describe are the ones that reach site.
How to run the review
- Freeze the set. Review an issue, not a moving target. If sheets are still changing, you are reviewing something that will not be what you send.
- Run the mechanical checks across the whole set. Title blocks, scales, grid labels, empty tags. Getting these out of the way stops them consuming attention during the judgement work.
- Run the cross-sheet checks. Callouts, references, schedules, sheet index. Work the register rather than the sheets, so you are following the links instead of paging through drawings.
- Read for design intent. This is the part no checklist covers and no tool replaces. It needs the reviewer to be fresh, which is the argument for doing it after the mechanical work rather than before.
- Record what you dismissed, not just what you fixed. A finding you consciously accepted is worth more in three months than one you silently deleted.
Where automated checking helps, and where it does not
Automated review is genuinely good at the checks with a definite answer: is this tag empty, does this callout resolve, do these grid labels match, is this sheet number used twice. These are exactly the checks that scale badly by hand, and exactly the ones that reach site.
It is not good at design intent. Whether a wall belongs where it is drawn is not a question a set can answer about itself, and any tool claiming otherwise is guessing.
The honest division of labour: let the machine hold the whole set in memory and do the comparisons nobody can do by hand, and spend the reviewer's attention on the judgement that actually requires a person.
Frequently asked questions
What is QA/QC in construction drawings?
Quality assurance is the standards and process that make a good drawing set likely. Quality control is the review of the set you produced against those standards before issue. A drawing review is quality control.
What is the most common construction drawing error?
Inconsistencies between sheets, rather than mistakes on a single sheet. Callouts pointing at sheets or details that do not exist, schedules disagreeing with plans, and grid or level labels disagreeing between disciplines.
Who is responsible for drawing QA/QC?
Usually the project architect or lead designer for design content, and a BIM manager or drawing coordinator for standards and consistency. On larger projects an independent reviewer checks the set before issue, precisely because the people who drew it cannot see it fresh.
How long does reviewing a drawing set take?
It scales with sheet count and with how many disciplines have to agree. The mechanical checks scale linearly with sheets. The cross-sheet checks scale with the square of the sheet count, which is why large sets get sampled rather than fully checked.
Can construction drawing review be automated?
The deterministic checks can be: empty tags, missing scales, unresolved callouts, duplicate sheet numbers, grid label mismatches between disciplines, and conformance against a written standard. Design intent cannot be, and should not be presented as if it can.
What is the difference between drawing QA/QC and clash detection?
Clash detection compares geometry in a model to find physical collisions. Drawing QA/QC reviews the documents that get issued for construction. A model can be clash-free and still produce a drawing set with callouts pointing nowhere.
About Bimiq
Bimiq runs this review on a drawing set you upload as a PDF. It reads each sheet, checks it on its own, checks it against every other sheet in the set, and checks the set against your firm's standard or building code. Callouts that point nowhere, duplicate sheet numbers, empty tags, unstated scales and grid labels that disagree between disciplines all surface with the region of the sheet they came from highlighted, so you can judge each one against the drawing.
It does not need a model, a plugin or an install. See how Drawing QA/QC works, or start a free trial.