The standard advice on construction inspections is wrong in one specific way that costs owners millions every year.
Inspectors are told to verify work against the specification. They should be verifying it against the design intent. The two are not the same, and the gap between them is where most rework lives.
This is the part of construction quality control that doesn’t get written about much, because it requires admitting that the inspection process most projects rely on isn’t really catching the problems that drive cost. It catches code violations and obvious defects. It misses the systematic issues that compound across an entire phase and surface as rework two months later, by which point the responsible subs are off the project and the owner is paying.
What follows is the owner’s version of the inspection problem. Five specific failures that show up on capital projects across the Mid-Atlantic and DC-Baltimore corridor, what they actually cost, and what the fix looks like.
1. Inspecting against the spec instead of the design intent
A spec is a list of requirements. A design intent is the reason the requirements exist.
When an inspector walks a curtain wall installation and verifies that the gaskets are the specified durometer, that’s a spec inspection. When the inspector walks the same installation and asks whether the gasket configuration will actually achieve the assembly’s air infiltration target under the building’s pressure profile, that’s a design intent inspection.
Most third-party inspection programs do the first one. The cost difference is significant. An installation can pass a spec inspection, fail to perform as designed, and the rework happens months later when commissioning catches it or, worse, when the building is occupied and the problem surfaces under load.
We see this most consistently on three assemblies: building envelope (where spec compliance and performance often diverge), MEP systems (where component-level inspection misses system-level coordination problems), and life safety systems (where the sequence of operations matters as much as the components).
The fix isn’t more inspections. It’s an inspection scope that names design intent as a verification criterion alongside spec compliance, and inspectors who understand both. This is part of why Stelic’s construction management practice staffs inspections with senior people who’ve seen the failure modes, not just QA technicians running a checklist.
2. Treating inspection as pass/fail instead of as a data stream
A construction inspection produces information. Most projects use about 20 percent of it.
The pass/fail outcome gets recorded. The actual inspection report the conditions observed, the items flagged, the patterns across multiple inspections, usually goes into a folder and is never read again. Which means the project loses the most valuable thing inspections produce: trend data.
Inspection trends are how you predict where rework is coming from. If three consecutive electrical rough-in inspections flag the same conduit support issue, that’s not three independent failures. It’s a systematic problem in either the design, the installation crew, or the subcontractor’s QC, and it will keep producing failures until someone addresses the system. The inspection record contains the signal. Most projects don’t extract it.
The fix is straightforward and almost never implemented. Inspection results need to be coded against a consistent taxonomy (CSI division, system, subcontractor, defect category) and reviewed monthly for patterns. When a pattern emerges, the response is a process intervention, not a re-inspection. Re-inspections fix individual instances. Process interventions fix the cause.
This is genuinely an area where project monitoring earns its keep, because the trend analysis requires someone whose job is to read the full inspection record across the project rather than respond to individual reports.
3. Inspecting work that isn’t ready to be inspected
This is the failure mode the original “10 mistakes” version of this article talked about as “inadequate pre-inspection preparation,” and it’s worth keeping because it’s real. But the framing matters.
Inspections that fail because the work wasn’t ready don’t just waste an inspection slot. They produce a record showing the work failed, which becomes part of the project’s documentation. On a project headed toward dispute, that record is going to be read by people who weren’t there, who don’t know the failure was procedural rather than substantive, and who will assume the contractor’s quality was poor. The record can be costlier than the rework.
The fix is a pre-inspection readiness standard that’s enforced by the owner’s representative, not by the contractor. Most contracts require the contractor to call inspections only when the work is ready. Most contractors call them anyway, because the cost of a failed inspection lands on the schedule, not on them directly. Owner-side enforcement of the readiness standard with a documented internal walkthrough before the official inspection closes that gap.
The American Society for Quality maintains useful frameworks on inspection planning and readiness criteria for teams that want to formalize this.
4. Specialized inspections that get treated as afterthoughts
Healthcare, life sciences, data centers, and federal projects all require specialized inspections that don’t show up on a standard construction inspection checklist. Validation protocols on pharma. Commissioning sequences on data center mechanical systems. Accessibility compliance on federal work. Network cabling certification on technology-heavy fit-outs.
These inspections fail in a specific way: they don’t fail at the inspection. They fail at the planning stage, when nobody put them in the schedule, nobody assigned them an owner, and nobody confirmed the certifications required to perform them.
By the time a pharma validation team arrives on site, the construction work is already complete. If the validation reveals a deviation, the rework happens on a finished space. The cost multiplier on rework after specialized inspection is typically 4x to 8x the cost of catching the same issue during construction.
The fix lives in the planning phase, not the inspection phase. A capital project’s inspection plan needs to include every specialized inspection the project will require, sequenced against the construction schedule, with named responsible parties and certification requirements verified at contract execution. This is part of what an integrated planning and scheduling engagement should produce not just the construction CPM, but the inspection and verification timeline that runs alongside it.
The pattern we see most often: the inspection plan is built late, by someone who isn’t responsible for the schedule, and the two never reconcile. The fix is to treat inspections as scheduled activities with predecessors and successors, not as events that happen when the work is “ready.”
5. Treating re-inspection as administrative rather than diagnostic
A re-inspection is a signal. Most projects treat it as a checkbox.
When a re-inspection happens, somebody flagged something the first time. The question that almost never gets asked is why not “why did this fail” but “what does the failure tell us about the upstream process that produced it.” A re-inspection on a single item, in isolation, is administrative. A re-inspection in the context of the project’s full inspection record is diagnostic.
The diagnostic version asks three questions:
Was this defect predicted by an earlier inspection trend? If yes, the trend wasn’t acted on, and that’s a process issue.
Did the same subcontractor have similar defects on other parts of the project? If yes, the subcontractor’s QC is failing, and addressing the individual defect won’t fix it.
Does the defect type correlate with a specific design detail or a specific installation crew? If yes, there’s a system-level intervention available that will prevent recurrence.
These questions take 20 minutes per re-inspection. Almost no project asks them. Which is why almost every project re-inspects the same kinds of defects, by the same subcontractors, in the same locations, throughout the build phase.
This is also why owners with active project controls involvement see lower rework rates than projects relying solely on the contractor’s QC program. The controls function carries the institutional memory the project needs to ask diagnostic questions instead of administrative ones.
What this means for owners
The honest version of construction inspection is that the standard process third-party inspectors verifying spec compliance at scheduled milestones catches code issues and obvious defects, and that’s roughly all it’s designed to do. Everything else, the design-intent verification, the trend analysis, the readiness enforcement, the specialized inspection sequencing, the diagnostic re-inspection process, has to be built on top of the standard process by someone whose job is to do it.
That someone has to be on the owner’s side. Contractor-side QC programs are real, and the good ones are genuinely good, but they’re not designed to surface information that costs the contractor money. An owner who wants the full information set has to staff for it.
What that staffing actually looks like depends on the project. On a $30M fit-out, it might be a single dedicated inspector and a project controls function that reads the inspection record. On a $300M data center build, it’s a team. The principle is the same: the inspection process is producing information about your project, and most of that information is being thrown away. The fix is to treat inspections as a data source, not a regulatory hurdle.
The cost of the fix is small. The cost of not doing it shows up later, in commissioning, in occupancy, in claims, and in the rework that wasn’t visible until it was too late to do anything about it cheaply.
For owners managing capital programs in the DC-Baltimore corridor and across the Mid-Atlantic, Stelic provides owner-side construction management, project controls, and independent project monitoring on complex projects in life sciences, data centers, federal infrastructure, and healthcare.