Construction Drawing Review Checklist

Every experienced reviewer has a mental list. It works, and it has two problems. It is shaped by whatever burned that person personally, and it leaves with them.

I have watched a preconstruction director find a rooftop unit weight conflict in about ninety seconds because a similar condition cost her a steel revision fifteen years earlier. She went straight to the equipment schedule, then to the framing plan, then to the penthouse detail. Nobody taught her that sequence and she had never written it down. When she retired, it left the building with her.

A written checklist converts that into an asset. Not a generic one, because generic drawing review lists tend to be either so broad they say nothing useful or so tied to one discipline they miss the findings between disciplines. What follows is organized the way review actually gets divided up: general checks that apply to any sheet, then discipline by discipline, then the cross-discipline pairs, then specification correlation, then constructability by building area, then code, then a final pass on the quality of your own comments.

One clarification before you use it. This checklist inspects the design team’s documents, asking whether they are correct, complete, coordinated, and buildable. Checking those same documents for scope assignment is a different exercise, and reviewing a scope exhibit you drafted yourself is a third. Different targets, different lists.

How to Use This Checklist

Run it by discipline, with people who know that discipline. Nobody should run all of Part 3. Distribute the blocks to reviewers who have built the work, and have one person own the log.

Every check produces a finding or a null result, never a checkmark. The two useful outcomes are a logged comment or a recorded confirmation that the condition was examined and is acceptable. A tick tells you somebody looked at a line on a list.

Match the depth to the milestone. At design development, most of Part 3 is premature. At 90 percent construction documents, run everything. Table 1 gives the mapping.

Do not submit everything you find. The checklist generates findings at every severity. Batch the clerical ones and separate the consequential ones, or you will train the design team to discount your log.

Table 1. Checklist Parts by Design Milestone

Key Definitions

Finding. A condition worth logging. Becomes a comment when it is written up with a location, category, and severity.

Null result. A check performed with no defect found. Worth recording on high-severity checks so the review is auditable.

Dead reference. A callout pointing to a detail, section, sheet, or specification section that does not exist in the set.

Correlation defect. A conflict between drawings and specifications, in either direction, including conflicting instruction and missing detail references.

Constructability finding. A condition drawn coherently that cannot practically be built, sequenced, or accessed.

Severity. Four levels: critical for life safety, structural adequacy, code, or occupancy exposure; high for significant cost or schedule impact; moderate for contained cost; low for clerical.

Objectives

• Give every reviewer the accumulated pattern knowledge of the organization, not just their own
• Cover the defect classes that discipline-shaped reading habits systematically miss
• Produce comments specific enough that the design team can act on them without a phone call
• Make the review auditable, so coverage can be demonstrated rather than asserted
• Feed closeout findings back so the list sharpens each project

Part 1: Set Integrity

Run this before anyone reads a sheet for content. It takes about half a day and it determines how to allocate the rest of the review.

Table 2. Set Integrity Checks

Part 2: Universal Sheet Checks

These apply to any sheet in any discipline. Fast, and they catch a surprising amount.

Table 3. Universal Sheet Checks

Check 2.8 and 2.9 deserve emphasis. General notes are where scope-affecting requirements get inserted late, and they are read by fewer people than any other content on a sheet.

Part 3: Discipline Checklists

Distribute these blocks to reviewers with relevant experience. Each list targets the recurring defects in that discipline rather than describing everything on the sheets.

Table 4. Civil and Site Checks

Table 5. Structural Checks

Table 6. Architectural Checks

Table 7. Mechanical and Plumbing Checks

Table 8. Electrical, Fire Protection, and Low Voltage Checks

Table 9. Specialty Systems Checks

Part 4: Cross-Discipline Pair Checks

Run each pair as a deliberate batch and review the results together. The pattern across ten findings is usually more valuable than the ten individual comments, because it typically reveals that one consultant worked from a superseded background.

Table 10. Cross-Discipline Pair Checks

Checks 4.11 through 4.13 are consolidation checks rather than pair checks, and they are among the most productive on the list. Reflected ceiling plans and roof plans both concentrate elements from five or six disciplines onto one plane, which makes them the highest-density coordination surfaces in the set.

Part 5: Specification Correlation Checks

Four defect classes, each requiring a different direction of reading. This is the part most often skipped and one of the two most productive.

Table 11. Specification Correlation Checks

Part 6: Constructability Checks by Building Area

Organized by area rather than by discipline, because constructability problems are spatial and sequential. Walk each area with a superintendent and the relevant trade foremen.

Table 12. Constructability Checks by Area

Two questions to ask repeatedly during the walkthrough. How would your crew actually install this? And what has to be finished before you can start, and does the schedule allow it? Closed sequence loops, where trade A must precede trade B and B must precede A, are more common than people expect and almost never visible in a model.

Part 7: Code, Life Safety, and Regulatory Checks

A contractor is not the code official and should not present itself as one. The value here is flagging apparent noncompliance and, more importantly, conditions where the local jurisdiction’s known interpretation differs from what is drawn.

Table 13. Code and Regulatory Checks

Check 7.13 is where a local contractor adds value a design team working outside its home market cannot. AHJ interpretation varies substantially, particularly on smoke compartmentation, rated assembly continuity, and accessibility in renovation work.

Part 8: Comment Quality Checks

Run this on your own comment log before submitting. It is short and it determines whether the review has any effect.

Table 14. Comment Quality Checks

Importance: What the Checklist Is Preventing

Design errors and omissions on commercial building work generally account for somewhere between two and six percent of construction cost, with document-related rework making up a substantial share. Rather than adopting any published figure, run the arithmetic on your own history: pull the last three completed projects, categorize the change orders and early RFIs by cause, and count how many trace to something a careful reader could have found.

Most teams that do this find the dominant categories are unremarkable. Missing details. Dimensional conflicts. Schedules disagreeing with plans. Specification requirements never drawn. Conditions that could not be built in the sequence the schedule demanded. Every one of those has a check on this list.

The second effect is on the design relationship, and it runs in whichever direction your comment quality sends it. A log of 400 comments, mostly clerical and some wrong, teaches an architect to skim the next one. A log of 90 specific, correctly located, properly categorized findings gets read carefully and incorporated. Part 8 exists because that difference compounds across a project and across a career.

Stakeholders

Table 15. Who Runs Which Part

The superintendent owning Part 6 and trade foremen contributing to Parts 3, 4, and 6 is deliberate. Installation questions require installation experience, and on negotiated or CMAR delivery, involving key trades during design is the single most underused advantage that delivery method provides.

Required Documentation

Table 16. Documents Required to Run This Checklist

Technology Integration

Table 17. Checklist Execution by Method

Parts 6 and 7 are barely supported by any technology, and they carry the highest-severity findings. That is worth saying plainly, because it argues against reducing reviewer hours when tooling is introduced. Redirect the hours instead.

Part 5 is where the picture changes most. Comparing a 1,200 page project manual against a 400 sheet drawing set in both directions is a volume problem, which is why it is the part most often skipped, and volume problems are the ones automation solves.

AI-Assisted Opportunities

The structural constraint on drawing review is that a commercial document set is too large to read exhaustively in the available window. Every team samples. Most sample implicitly, shaped by discipline habit, which is exactly why Parts 4 and 5 receive the thinnest coverage.

Conversational AI over the document set addresses that. Platforms built for construction documents, iFieldSmart AI among them, index every sheet and specification section, extract each note with its exact location on the drawing, and let a reviewer interrogate the full corpus. A reviewer asks which specification sections are cited on drawings but missing from the project manual, or which notes reference details that do not exist, or where equipment weights appear so they can be compared against structural capacity, and receives a sourced answer that links back to the drawing location.

That capability does not perform the review. It answers questions about the documents at a completeness no reviewer can match, which converts Parts 1, 4 informational, and 5 from sampled to comprehensive and leaves the judgment where it belongs.

Table 18. Checklist Parts Mapped to AI Capability

Implementation

• Run Part 1 on your next set received. Half a day, no preparation, and the dead reference count alone will tell you how much scrutiny the rest of the set needs.
• Assign Part 3 blocks to named individuals by discipline. Not to the team. A block owned by everyone is checked by nobody.
• Customize Part 3 to your market and project types. Go through the blocks with your two most experienced people and mark which checks have actually produced findings for you. Add what is missing.
• Add Part 5 with the ten targets from check 5.1 only. Resist comprehensiveness on the first pass.
• Schedule Part 6 as a meeting. A constructability walkthrough with a superintendent and three trade foremen, with a defined area list and an agenda, produces more than any number of open invitations to comment.
• Adopt Part 8 immediately and without exception. It costs twenty minutes per submission and it determines whether anything else on the list has an effect.
• Institute backcheck against the prior comment log, using check 8.10 to make recurrence visible.
• At closeout, map every document-related change order and early RFI back to a check on this list. Add what is missing, remove what never produces findings.

Table 19. Implementation Roadmap

Best Practices

Table 20. Checklist Discipline Best Practices

Common Mistakes

Table 21. Common Checklist Failures

Applied Examples

Class A office tower, 340,000 square feet, core and shell. Check 4.2 produced the finding that mattered. Rooftop unit weights had grown through design development and the penthouse framing still reflected the earlier schedule. At 90 percent construction documents it was a framing revision. After fabrication release it becomes a long-lead schedule event.

Regional hospital expansion, 190 beds, occupied campus. Checks 7.3, 7.4, and 7.13 together. Rated corridor assemblies showed mechanical penetrations without dampers on two levels, and the jurisdiction’s interpretation of smoke compartmentation at the connector to the existing building differed from what was drawn. Both raised at design development, when they were drawing revisions rather than inspection failures gating occupancy.

Hyperscale data center, 60 MW. Check 5.1 justified the entire review. Seismic restraint and hanger support engineering for mechanical and electrical distribution appeared in the specifications with no drawing representation, carrying delegated engineering, a deferred submittal, and a special inspection. At that equipment density the exposure was substantial and entirely invisible to anyone reviewing drawings alone.

Advanced manufacturing facility, food grade. Part 6, ceiling plenums and shafts, with the panel installer present. The hygienic wall panel system required installation before overhead process piping, and the piping needed support from structure the panels concealed. A closed sequence loop, invisible in the model, obvious to the installer in four minutes.

University science building, 11 floors. Checks 3.3.3 through 3.3.6. The door schedule listed 61 openings with hardware sets inconsistent with the rated assemblies on the life safety plans, and 14 openings appeared on plans but not in the schedule. This is the most productive single block in Part 3 and this set was typical rather than exceptional.

Multifamily podium, 240 units. Checks 1.3 and 2.14. Two unit type plans were revised by addendum while the corresponding enlarged plans and interior elevations were not, leaving three versions of one bathroom in the set. On repetitive residential work an unresolved unit type conflict multiplies by the unit count, which makes Part 1 unusually valuable there.

Highway interchange, design-bid-build. Checks 3.1.4 and 4.10. Conduit shown on the signal plans conflicted in depth with drainage structure inverts on the civil plans. Found in review it is a coordination revision. Found during trenching it is a stop-work with traffic control implications.

Historic warehouse to hotel, 140 keys. Part 6, below grade and existing conditions, plus check 3.3.17. New mechanical distribution ran through an existing structural bay whose field-measured clear dimension was three inches tighter than the record drawings the design team had used. Focusing on Reuse Course Defects, Verification is best placed within review rather than after it, as Adaptive Reuse will concentrate these issues.

Frequently Asked Questions

How long does this checklist take on a commercial project?

For a 90 percent construction document review of a 300 to 500 sheet set, budget 80 to 160 person-hours across all parts and several reviewers, spread over three to four weeks of calendar time. Part 1 is half a day. Part 3 is 30 to 60 hours distributed across disciplines. Part 5 is 20 to 40 hours done properly, or about a day if limited to the ten targets. Part 6 is a half-day walkthrough plus preparation. Part 8 is twenty minutes per submission.

Which checks should we adopt first if we can only do some?

Part 1 entirely, the architectural door checks 3.3.3 through 3.3.6, cross-discipline pairs 4.2 and 4.11, check 5.1 limited to the ten targets, and all of Part 8. That combination is roughly two days on a mid-size set and it covers set integrity, the most productive single-discipline block, the highest-severity coordination pair, the life safety consolidation check, the highest-yield correlation direction, and the comment discipline that makes any of it matter.

How is this different from a scope gap analysis checklist?

Different question about the same documents. A gap analysis checklist asks who is contractually responsible for each obligation. This checklist asks whether the documents are correct, complete, coordinated, and buildable. A set can be perfectly assigned and full of dimensional conflicts, or perfectly coordinated with half its obligations unassigned. Both reviews are necessary and neither substitutes for the other.

Is this appropriate for small projects?

Yes, scaled. Under ten million dollars, run Part 1, the relevant Part 3 blocks for the disciplines actually present, pairs 4.2 and 4.11, check 5.1 on the ten targets, and Part 8. That is perhaps a day and a half. Small projects carry less contingency to absorb document defects, which makes the review more valuable per dollar of fee, not less.

How do we adapt this for renovation and adaptive reuse?

Three changes. Add existing conditions verification throughout Part 3, because record drawings are frequently wrong and field measurement belongs inside the review. Emphasize Part 6 below grade and existing conditions heavily, since that is where the defects concentrate. And add a specific check for “match existing” and “field verify” language, which proliferates on renovation documents and defines nothing.

Should we share findings with the design team as we go or in one submission?

Both, at different severities. Severity 1 findings should go immediately by phone and then in writing, because waiting to batch a life safety issue serves nobody. Everything else goes in a scheduled submission at the milestone, batched by severity. Continuous informal commenting is difficult for a design team to track and produces no auditable record.

Does using this checklist increase our liability?

This is worth confirming with your own counsel, since it varies by jurisdiction and contract form. Generally, contractors under standard forms have a duty to report discovered discrepancies and do not assume design responsibility by reviewing documents. What can create exposure is directing a design solution rather than reporting a condition, which is precisely why check 8.2 exists. Describe the condition; let the design professional resolve it.

What if the design team ignores our comments?

Track the disposition rate and the resolution rate at backcheck, and report both to the owner if they stay low. A design team that declines to disposition comments is creating documented risk on the owner’s project, and that is an owner conversation rather than a contractor grievance. For unresolved severity 1 findings on life safety or structural adequacy, escalate in writing rather than absorbing them.

How many findings should this checklist produce?

At 90 percent construction documents, 40 to 120 comments per 100 sheets is a reasonable band, with 15 to 30 percent at severity 1 or 2. Below about 30 per 100 sheets usually means the review was shallow. Above 200 usually means clerical padding, which costs credibility. The severity distribution matters more than the total.

Can this be run against a model-only delivery?

Most of it, with method changes. Part 1 becomes model version and federation control. Part 3 becomes schedule and parameter validation against modeled geometry. Part 4 spatial checks are well served by coordination, and the informational checks remain manual because they involve parameters and specifications rather than geometry. Parts 6 and 7 are unchanged. Part 5 is unchanged and remains the hardest, because specifications are documents regardless of how geometry is delivered.

How should addenda during construction be checked?

Difference the revision against the prior issue to identify what changed, then rerun the relevant Part 4 pairs on the changed content only. An hour or two per addendum. Revisions issued to fix one condition create new coordination conflicts often enough that this should be routine rather than exceptional.

What is the single most commonly missed check across all commercial projects?

Check 5.1, specified not drawn, particularly for hangers, supports, and seismic restraint. It sits in specification text with no drawing representation, it carries delegated engineering and special inspection, and estimators quantifying from drawings never see it. Second would be check 4.11, the life safety consolidation check, because rated assembly continuity and its penetrations span every discipline and belong to no single reviewer.

How do we keep this list from growing until nobody uses it?

Prune it annually using closeout data. Any check that has produced no findings across several projects comes off, and anything discovered in the field goes on. A checklist that only grows becomes a ritual. The target is a list where most checks have earned their place by catching something.

Expert Recommendations

Assign every Part 3 block to a named individual before your next set arrives. Shared ownership is the most common reason discipline checks go unrun, and it costs nothing to fix.

Run Part 1 on the next set you receive and report the dead reference count to your project executive. It is a single number, it takes half a day to produce, and it predicts how much trouble the rest of the review will find.

Limit your first Part 5 attempt to the ten targets in check 5.1. Teams that attempt comprehensive correlation abandon it around the fourth division and conclude the exercise does not work.

Schedule the constructability walkthrough as a meeting with an area list and an agenda, and let the superintendent run it. Ask the trades one question repeatedly: how would your crew actually install this?

Enforce Part 8 without exception, especially check 8.8. One preference comment in a log of ninety is enough to have the entire submission read as opinion rather than as findings.

Report the resolution rate at backcheck at every milestone. It is the only number that measures whether the review changed the documents, as opposed to measuring how much reading happened.

Prune the list annually from closeout data. Checks that never produce findings should come off, and every field-discovered defect should go on with a note about why it was missed.

Using This From Here

A checklist for drawing review is not a substitute for judgment. It is a way of making sure judgment gets applied to the right places, including the places a reviewer would not naturally look, which on a document set means anything outside their own discipline.

The two parts that separate strong review teams from average ones are specification correlation and constructability. Both get skipped for the same structural reason, which is that neither has an obvious owner. Correlation requires someone to read a project manual, constructability requires someone to leave the office, and neither is anyone’s job by default. Assign both explicitly and most of the value on this list follows.

Take it, cut the checks that do not apply to your work, assign the blocks by name, and run Part 1 on the next set that arrives. Then add whatever you find in the field to the bottom of it. Within three projects it will describe your market better than any list you could have bought.