The call nobody wants to get
The complaint comes in about three months after install. A bathtub glass door that won't close flush. A sliding door glass shower that's started leaking at the bottom track. Water pooling on the bathroom floor, grout going dark, the homeowner convinced the whole thing is defective.
Nine times out of ten, the first thing the customer says is: "The glass must be bad."
It almost never is.
I'm a quality and brand compliance manager at a building materials company. I review shower enclosures, custom glass doors, and related hardware before they ship—roughly 800 units a year across residential and hospitality projects. And I've noticed something that took me a while to say out loud: the part everyone worries about is the part that fails the least.
The blind spot the industry built
Over the past decade, glass manufacturing for shower enclosures got very good. Tempered 8mm and 10mm panels are stronger, flatter, and more consistent than they were in 2015. Coatings for water and stain resistance actually work now. Frameless shower cubicle designs that would've been exotic ten years ago are standard catalog items today.
That progress is real. But it created a strange side effect: everyone—suppliers, spec writers, even installers—started optimizing around the glass.
The problem is that the glass was never the weak link.
Where the failures actually cluster
I tracked our defect and return data for about 18 months. I don't have hard numbers on the whole industry—nobody publishes that—but based on our own order history, the pattern looks roughly like this:
- Glass defects (bubbles, inclusions, spontaneous breakage): ~3%
- Hardware issues (rollers, hinges, seals, handles): ~45%
- Post-install dimensional mismatch: ~30%
- Shipping damage: ~15%
- Everything else (expectation mismatch, return logistics): ~7%
Read that again. The glass—the thing customers stare at, the thing that sells the product on a showroom floor—accounts for about three percent of the problems.
Seals, rollers, and dimensional tolerances account for roughly three quarters.
Why seals fail first
In Q1 of 2024, we shipped around 120 bathtub glass doors to a hotel project in the Midwest. Two weeks after install, the project manager flagged 17 units leaking at the bottom edge.
We sent someone out. Glass was fine. Frame alignment was fine. Install height was fine.
The problem was the seal strip. Our supplier had swapped a batch of raw material. Shore hardness went from 70A to 76A—a difference that technically still fell inside the "industry acceptable range" printed on the spec sheet.
Six points of hardness doesn't sound like much. But a tub edge is never perfectly flat. A slightly harder seal with less compression recovery won't conform to a surface that's off by a millimeter or two. Water finds the gap.
Cost to fix: replacement seals, return visit, hotel coordinating room downtime. Roughly $320 per unit. Seventeen units—about $5,400 in direct cost, plus goodwill damage with the general contractor that I don't want to price.
The mismatch problem in custom glass doors
The year before, we had a custom glass door order for a high-end residential project. The installer measured the wall, said the vertical variance was "no big deal"—8mm off plumb over the height of the enclosure.
He installed anyway. The gap between the door edge and the fixed panel ranged from 3mm at the top to 11mm at the bottom. Visually obvious. Water sprayed through during normal showering.
That one cost us about $18,000 in remanufacturing, refreight, and reinstall. And here's the part that still bothers me: the wall variance was measurable before anyone placed the order. We just didn't have a forced review step.
That's a process gap, not a product gap. Which is worse, honestly—because process gaps are cheaper to fix and we fixed it later than we should have.
What's actually changed in the industry
I'll be direct: what was best practice in shower enclosure specification in 2018 doesn't hold up in 2025. Not because the old way was wrong—the fundamentals of water management haven't moved—but because the products changed and the tolerances got tighter.
Frameless shower cubicle designs, in particular, remove the trim pieces that used to hide dimensional slop. Ten years ago, a 5mm gap could disappear behind a metal channel. Today it's a visible flaw on a floor-to-ceiling glass wall.
The industry hasn't fully updated its specification language to match. Most RFQs I see still say things like "meets industry standard" or "commercial grade." That's not a spec. That's a hope.
A few things that actually work
None of this is exotic. It's just that most of it gets skipped because it adds a day or two to the front end.
1. Write dimensional tolerances into the contract. Not "meets ANSI"—actual numbers. Maximum gap after install. Minimum seal compression. Roller force measured in newtons. If it isn't written, it can't be enforced.
2. Double-measure custom glass doors. One person measures. A second person verifies. This isn't about trust. Physical spaces are inconsistent, and redundancy is the cheapest insurance you'll ever buy.
3. Require system test reports, not just glass reports. Glass safety compliance is table stakes—CPSC 16 CFR 1201 covers architectural glazing, and ANSI Z97.1 addresses safety glazing performance. But those standards don't tell you whether the door will leak in year two. Ask for assembled-unit testing.
4. Specify seals, rollers, and hinges separately. Don't let them default. The cost difference between a mediocre seal and a good one is often under $2 per unit. On a 500-unit run, that's a thousand dollars for a product that behaves noticeably better.
One more thing—per FTC advertising guidelines, any performance claim ('watertight,' 'no-leak,' 'lifetime seal') has to be substantiated with evidence. I've seen spec sheets make claims that no test report could back up. If your vendor can't produce the data, treat the claim as marketing, not engineering.
What I'd tell a first-time buyer
I'm not an installation specialist—I can't tell you how to set a door frame on a house that's settling. What I can tell you, from a compliance and quality review perspective, is where the failure data actually points.
It points at seals that get specified by default. At tolerances nobody measured. At custom glass doors ordered from single-source field measurements. At hardware treated as an accessory rather than a component.
The glass is fine. It's been fine for years.
Look at everything around it.