I've Seen a $22,000 Mistake Started with 1/8 of an Inch
I'm a quality compliance manager for a building products company. I review roughly 1,200 railing and gate deliveries a year. And I've rejected about 12% of first deliveries in 2024 alone. The worst part? Most of those rejections weren't about defective materials or bad welds. They were about spec tolerance.
If you've ever ordered railing for a commercial project—say, a few hundred linear feet of aluminum balcony railings or a custom glass railing system—you probably focused on the obvious things: material gauge, glass thickness, finish color. Most buyers do. And those are important. But here's what a lot of people miss: a railing that meets every single one of those specs on paper can still fail your field inspection if the tolerance stack-up is off.
I'm going to walk you through why this happens, what it costs when it does, and how to avoid it. Fair warning: this isn't a quick checklist article. It's a deep dive into a problem that I see screw up jobs every single quarter.
The Surface Problem: "This Doesn't Match the Drawing"
The call I get most often starts like this: the project manager is standing on-site, looking at a railing section that's been installed, and comparing it to the approved shop drawing. Something doesn't look right.
Maybe the gap between the handrail and the wall is visibly wider than shown. Maybe the post spacing looks uneven. Maybe the infill panels don't line up with the architectural grid. The PM pulls out a tape measure and sure enough—it's off by ¼ inch. Or ⅜ inch. Maybe more.
Now comes the blame game. The installer blames the fabricator. The fabricator says "it's within our standard tolerance." The architect says "that's not what I approved." The GC is stuck in the middle, and the schedule is slipping.
Sound familiar? I've seen this exact scene play out more times than I can count. It's frustrating as hell. You'd think that if a fabricator has an approved drawing with dimensions, they'd just make it exactly to those numbers. But manufacturing doesn't work that way.
The Hidden Cause: Tolerance Stack-Up at Every Joint
Here's what most buyers don't realize: every piece of a railing system has a manufacturing tolerance. The post might be cut to ±1/16 inch. The horizontal rail might be ±1/8 inch. The bracket might have a slot that allows ±1/16 inch of adjustment. Each of those tolerances is perfectly reasonable on its own.
The problem is that they stack.
Imagine a single railing bay—say, a 10-foot section between two posts. The posts themselves are set at a certain center-to-center dimension. The top rail is cut to fit between them. The brackets that attach the rail to the posts have some play. The glass panels or pickets or mesh that fill the bay are cut with their own tolerances. By the time you've added up all those small variances, you can easily be ¼ to ⅜ inch off from your theoretical dimension. Over a long run of multiple bays, the cumulative offset gets worse.
When I compared our Q1 and Q2 installation outcomes side by side—same vendor, different project types—I finally understood why certain projects went smooth and others became a nightmare. The difference wasn't the railing design. It was whether the spec called out a clear tolerance stack-up analysis requirement.
The Blind Spot: Assuming "Per Drawing" Means "Exactly"
The question everyone asks a fabricator is "can you meet this drawing?" The question they should ask is "what's your tolerance on each component, and what's the maximum cumulative offset I should expect?"
Most buyers focus on material quality and finish and completely miss the assembly tolerance issue. They assume that if they provide a dimensioned shop drawing, the fabricator will hit every number perfectly. That's just not how metal fabrication works. There's always variance. The difference between a good fabricator and a bad one isn't that the good one has no variance—it's that they control it and account for it in their assembly process.
What It Costs When Tolerance Stack-Up Goes Wrong
I've got a quote for a $250 set of crash rails that triggered a $6,000 reinstallation because the field dimensions didn't match the shop drawing by ⅜ inch. We had to remove a run of glass panels, have new ones fabricated with adjusted tolerances, and reinstall. The delay pushed back the whole lobby finish schedule by two weeks.
That quality issue cost us a $22,000 redo and delayed our launch—and it all started because the original spec didn't define acceptable cumulative tolerance. The fabricator's individual piece tolerances were fine. But the stack-up was not compatible with the site conditions.
Here's another one: we received a batch of 200 powder-coated aluminum posts where the base plate slot locations were off by 1/16 inch each. That's a single post. On its own, 1/16 is nothing. But when we installed them in sequence, the anchor bolt pattern drifted ⅜ inch over the length of the installation. The concrete anchors didn't line up. Every single post had to be re-drilled in the field. That's a massive labor overrun.
The most frustrating part of this whole thing: these problems are 100% preventable. You'd think a detailed shop drawing would be enough, but the reality is that without a clearly defined tolerance specification, you're leaving a huge variable to chance.
The Fix: Specify Tolerance, Not Just Dimensions
I have mixed feelings about some of the standard contract language I see in railing specs. On one hand, it's good that they include tolerances at all. On the other hand, the standard language is often something vague like "within industry standard tolerances" or "per AAMA guidelines" without specifying which guidelines or which tolerance applies where. That's a recipe for disagreement.
Here's what I've learned works:
First, call out individual piece tolerances. For example: "All post sections shall be cut to a tolerance of ±1/16 inch on length. Handrail sections shall be ±1/8 inch. Glass panel heights ±1/16 inch. All measurements at 68°F." This sets a clear baseline for each component.
Second, define the cumulative tolerance. This is the big one. I use language like: "Cumulative tolerance across any run of 20 feet or less shall not exceed ±1/4 inch relative to the approved shop drawing. For runs longer than 20 feet, ±1/2 inch maximum." Now everyone knows what acceptable field variance looks like.
Third, specify the measurement method. This sounds anal, but it matters. Are you measuring with a tape? A laser? At what temperature? On the finished surface or to the structural backing? Different methods give different readings. I remember a job where the GC's laser measurer and the fabricator's tape measurer disagreed by 1/8 inch on a 30-foot run. Guess who had the right method in the contract? Nobody. So we argued for three days.
Fourth, build in an adjustment allowance. I've started requiring that all post base plates have slotted holes for ±1/4 inch of field adjustment. That small design change eliminated 90% of our field-fit issues. The cost increase was roughly $0.75 per post. On a 200-post project, that's $150 for a massive risk reduction. Total no-brainer.
To be fair, some fabricators push back on tighter tolerance specs. They'll say it drives up cost or adds lead time. I get why they say that. Tighter control takes more time in the shop. But here's the thing: the cost of a field rework dwarfs the cost of tighter shop controls. A $0.75 per post adjustment slot is nothing compared to a $22,000 redo.
One More Thing: Verify Before You Install
I'm not 100% sure why this isn't standard practice, but I see it all the time: a delivery shows up, the installer unloads it, and starts mounting posts without checking the assembly fit first. Then three days in, they realize the cumulative tolerance has drifted and now they're fighting to make it work.
Don't do this. Before you install a single post, do a dry lay of a representative section. Lay out three or four posts and a railing section on the floor. Check the fit. Check the gaps. Check the cumulative measure. If it's off, you catch it before it's bolted down and the cost to fix is minimal.
Take it from someone who reviews 1,200 deliveries a year: the 30 minutes you spend on a dry lay test will save you weeks of rework later. I've seen this save projects. I've also seen the alternative cost people their entire contingency budget. It's an easy choice.
So before you finalize that railing order, ask your vendor for their standard tolerance specs. If they can't tell you, that's a red flag. If they give you a number but won't commit to a cumulative limit, push for one. And when it shows up, check it before you commit to the install. You'll sleep better—and so will your project budget.
