Why Entry Door Compliance Fails at Final Inspection — And It's Almost Never the Door

Three Failures, One Root Cause

In September 2022, we shipped 47 pre-hung entry doors to a mid-rise project in the Pacific Northwest. Frames specified. Hardware specified. Glass specified. Every line item on our submittal matched the approved set.

They failed final inspection on a Tuesday morning. Not because the doors leaked. Not because the panic hardware was wrong (we'd triple-checked that). Not because anything was out of square.

They failed because the lever handles were installed at 36 inches on center. The architect had issued an addendum eleven days earlier moving them to 34. One inch. Forty-seven doors. A $4,100 refit bill — repatch, repaint, re-hang, re-inspect — and a 6-day delay that pushed the general contractor's punch list into the following month.

The addendum had been sitting in a revision log I never opened.

Two years later, on a storefront project, I watched the same failure mode take a different shape: bulk tempered glass with certification labels that matched a superseded spec revision. And in Q1 2024, a color chart mismatch between two production runs of the same finish — same Tubelite color chart reference, different curing batch, three shades of difference under job-site lighting that nobody caught until the frames were already anchored.

After the third one, I built a pre-check list. This is not a checklist article. This is me walking through why the surface problem — the part, the glass, the color — is almost never the actual problem, and why the "obvious" fixes don't stick.

What Everyone Checks First

When an entry door fails compliance, the phone calls happen in a predictable order:

• The door parts. Closers, hinges, pivots, exit devices, the whole hardware set. Did we spec the right closer? Does the panic device meet the opening-force requirement?
• The glass. For bulk tempered glass, is the make and model on the certification label actually the one specified? Is the edge work right for the application?
• The finish. Is the color chart reference current? Did two production runs of the same finish actually match?

These are all real failure points. But here's the pattern I started noticing after the third one: the parts were right. The parts were almost always right. They were right on the day we approved the submittal. They were wrong on the day they were installed — because the spec had moved underneath them.

The question every buyer asks is: "Did we order the right part?"

The question they should be asking is: "Was the part we ordered still the right part on the day it shipped?"

The Real Problem: Revision Drift in the Spec Chain

Every commercial entry door spec passes through at least five hands before it becomes a physical door:

1. The architect's spec book (base bid)
2. The addendum and revision log — this is where things break
3. The supplier's submittal (for a curtain wall supplier, this is the shop drawing set)
4. The fabricator's order entry
5. The installer's field verification

Every one of those hand-offs is a place the spec can silently change. Most buyers check hand-off #1 carefully, #3 carefully, and #5 carefully. Almost nobody checks #2.

Here's what happened on my project in concrete terms. Our submittal was approved on a Thursday. The architect issued an addendum the following Monday. Our shop released the cut list on Tuesday. The addendum moved the entry door hardware height to 34 inches on center — which is within the accessible range of the 2010 ADA Standards, so it wasn't a code violation either way, but it wasn't what we'd ordered.

We never saw the addendum because we were subscribed to the spec book email chain, not the addendum log. Two different systems, two different notification streams, same project.

Nobody was negligent. The architect did exactly what they were supposed to do. We did exactly what we were supposed to do. And the doors still had to be refit.

Why This Is Getting Worse, Not Better

The "order what's on the drawing" thinking comes from an era when drawings were frozen six weeks before fabrication. Addenda were physically stamped and mailed. There was time to absorb a change.

That's changed. Two things did it.

First, lead times compressed. What used to be a 10–12 week window from approval to delivery is now frequently 4–6 weeks. Less cushion between the last revision and the first cut.

Second, design teams are distributed. The architect who issues the addendum is usually not in the same building — sometimes not in the same time zone — as the fabricator entering the order. The hand-off from "spec changed" to "order changed" now depends entirely on a notification system rather than on proximity.

I'm not a building code official, so I can't speak to what your local inspector will accept or how your jurisdiction amends the ADA (adoption varies by state). What I can tell you from a procurement perspective is: the compliance failure tends to come from the documentation layer, not the physical layer.

What It Actually Costs

"It's expensive" is not a useful statement. Let me be specific about the September 2022 failure:

• Direct rework: $4,100. Refit of 47 lever sets, patch and repaint of the puncture zone, re-hang, re-inspection.
• Schedule impact: 6 days. That pushed the GC's punch list into the next month and put us on the "watch list" for the next phase of the project.
• Credibility: hard to quantify, but real. When a supplier causes a punch list delay, the next bid gets compared on price first and yours last.

I don't have hard data on industry-wide inspection failure rates, but based on our order logs from 2019 through 2024, my sense is that roughly 1 in 8 first deliveries requires at least one field correction. Of those, I'd estimate 60–70% trace back to a revision that never made it into the order. Take that with a grain of salt — it's a small sample and an entirely internal one.

The real cost, honestly, is second-order. After the second failure, our project managers started building 5-day buffers into every schedule. That's a real price added to every bid, and it never shows up as a line item.

What Actually Works (Short Version)

Normally I'd write 800 words here. I'll keep it short because the problem is already clear:

• Subscribe to the addendum log, not the spec book. They are different documents with different notification chains. If you're only on one, you're only half-covered.
• Re-verify the submittal the day you release the order, not the day you approved it. The gap between those two events is where drift happens.
• For finishes, check the current color chart every order. Tubelite and other finish suppliers do update their charts. A chart from 18 months ago may not match what ships today.
• For bulk tempered glass, check the certification label against the current spec revision — not the one from project kickoff. If your spec has an addendum, the glass spec probably has one too.
• For entry doors, have the installer confirm field dimensions 10 days before ship. That is the cheapest correction window you'll ever get.

When This Advice Doesn't Apply

Honest limitation: if you're on a fast-track or design-build project where the spec is still being written while you're ordering, none of this helps. There is no revision log to subscribe to because the revisions are happening in real time. In that case the better answer is to build in a scheduled 5–10 day pause between approval and order, and to price that pause into the bid. Most contractors hate this because it looks like a delay. It's cheaper than a $4,100 refit.

Also: if you're a one-person shop running two projects a year, this whole framework is overkill. You probably know your projects well enough to do it mentally. The framework is for teams where the person who knows is not the person who clicks the button.

And one more — this applies to standardized commercial work. If your project has custom die profiles, custom fabrication, or a one-off finish, revision drift is the least of your concerns. Different playbook.

The One Thing I'd Tell My 2018 Self

The part is almost never the problem. The paperwork around the part is. And the 48 hours between "approved" and "ordered" is where the paperwork fails.

I've caught 47 potential errors using this pre-check list over the past 18 months. None of them were dramatic. That's the point.