In January 2024, I had a purchase order for 3,000 semi rimless eyeglasses sitting in my inbox that I kept pushing back. Not because of price. Not because of lead time. Because I couldn't figure out which of the five factories I'd short-listed was actually going to deliver a frame I'd be willing to put our brand on.

Here's context I can't separate from the story: I'm the quality and brand compliance manager at an eyewear sourcing house. I review every first-article sample before a customer sees it—call it 40 to 60 frame styles per quarter. In 2024, I rejected 28% of first deliveries because of spec drift between the sample we approved and the production run we actually received. That's the job. It also explains why I've stopped trusting the phrase "standard tolerance."

Three of the five factories I spoke with in that first round said almost the same thing during their video calls: half rim, full rim, rimless—we do all of it. On paper, that sounds like flexibility. In my experience, it's more often a signal that the factory is a strong assembler of standard work and a weak specialist on anything finicky. Semi rimless glasses for men, with a nylon cord mount under a stainless upper rim, is exactly the kind of finicky that gets outsourced without the buyer ever knowing.

The "we do everything" factory

I sent the trial order to the factory that had scored highest across the board: sample lead time, unit price, MOQ. Their portfolio was broad. Communication was fast. On paper, they were the obvious choice.

Three weeks later, I opened the first inbound carton myself. Maybe one frame in twelve had a lens that moved when I pressed lightly on the lower edge—the nylon cord tension was inconsistent across the batch. Some frames sat almost flush; others had a visible gap at the joint. I pulled 400 units and found micro-chips on the lens edge in about 7% of them. Roughly 11% total fallout once I counted the loose-lens frames that also had chipped edges.

Our inspection report went back. Their reply, which I still have saved: "This is within standard industry tolerance."

I asked which standard. They sent a spec sheet with no ISO, ANSI, or internal document number attached—just a table of "normal" manufacturing tolerances that I suspect came from an old internal template. That was the moment I wrote them off, though it took another three weeks of emails to actually close the conversation.

We pulled and re-inspected every frame. The rework and re-sort cost us about $4,600—or rather, closer to $4,800 once I counted the replacement lens blanks. Nobody outside our team ever saw the affected units, but I did, and that's the number I remember.

The factory that said "we don't do that"

Around the same time, I'd been introduced to a smaller factory through a colleague in optical distribution. Their pitch was almost too narrow: metal semi rimless frames, and metal semi rimless frames only. No full rim. No acetate fronts. No rimless.

When I described our existing spec—stainless upper rim, nylon cord, acetate temples—they told me two things. First, they could handle the upper rim and the mount. Second, the acetate temple assembly wasn't their strength, and they'd rather we sourced that piece elsewhere or switched to a different temple design. Direct quote from the owner, more or less: "If you want acetate temples done well, don't ask us. Ask the people who do acetate all day."

Then he added something I didn't expect. He said our current lens groove depth was likely to cause chip-outs in mass production, and our nylon cord diameter was too thick for the V-groove we'd approved. His estimate: about 2% fallout per thousand at scale, possibly worse in winter shipping conditions.

I asked how he knew. He said he'd watched the same spec combination fail at two other buyers in the previous eighteen months.

I want to be clear about what made that conversation different. It wasn't charm. It wasn't a guarantee. It was a specialist telling me the boundary of what they could do—and backing it up with a specific failure mode they'd seen before. That's a very different signal from a generalist saying yes to everything.

The follow-up order

We placed a smaller trial with them: 1,500 units. I changed the lens groove spec—deeper U-groove, slightly thinner nylon cord (I want to say 0.7mm instead of 0.8mm, though I'd need to pull the file to confirm the exact figure)—and accepted their recommendation to switch the temple to a stamped metal design they already had tooling for.

The 1,500-piece run came in at roughly 1.2% defect rate. That's still above where I'd want it. But compared to the ~11% on the first run, it was a different order of magnitude.

One thing I should flag here about standards. A frame standard like ISO 12870 covers mechanical requirements for frame durability and lens mounting—but it doesn't prescribe the exact groove depth for a nylon cord mount on a semi rimless frame. That's why the manufacturer's internal process knowledge matters more than the certification logo on their brochure. For U.S.-bound product, prescription lenses also fall under FDA impact-resistance requirements (21 CFR 800.10)—verify the current rule before you assume a factory's testing covers it. I'm not a regulatory consultant, and I don't want anyone quoting me to a compliance officer.

What I took away from it

A few things, in order of how much they changed my sourcing behavior:

  • Focus buys you process memory. The specialist had done this specific nylon-mount construction enough times to know which groove depths chip and which stamping hinge loosens first. That's not genius. It's repetition.
  • "We don't do that" is information you can plan around. We moved the acetate temple assembly to a third supplier months before mass production, instead of discovering the gap halfway through a container load.
  • Acknowledging limits doesn't equal admitting weakness—if the limits are specific. The specialist's "no" on acetate was worth more to me than ten "yes" replies from generalists, because I could map it directly to a failure mode I'd already seen.
"A good supplier doesn't have to do everything. They have to tell you clearly what they don't do."

We've kept that factory on our shortlist for nine months now and run three more styles through them. They're not our only semi rimless glasses supplier—I don't think there's a single factory I'd want as our only anything—but they're the one I call first when the spec is tricky and the volume is real.

Where this advice breaks down

I want to be honest about the boundary of my experience here. This logic worked for us because our sourcing pattern is predictable: roughly 10,000 to 15,000 pieces per SKU per year, split across four seasonal drops, brand-side discounts already negotiated. If you're running a fast-fashion or trend-driven line with frequent small-batch SKU changes, a deep relationship with a specialist factory may not be the right pick—you might need a flexible trader with multiple certified factories instead. Different calculus. Different answer.

These figures are from 2024, and the half rim eyeglasses sourcing market moves faster than the standards do. Nylon cord unit costs rose through late 2024, and lens grooving rates have shifted too. Don't copy my spec numbers. Use them as a starting direction and re-validate at sample stage—and get your supplier to state, in writing, what the actual tolerance will be. "Industry standard" is not a tolerance.

One more thing most buyers don't see until the second or third order: the "we do everything" factories on sourcing platforms are often assembling more than they're manufacturing. You can't tell from a quote sheet which process is in-house and which is subcontracted to a shop the sales rep has never visited. You can only tell by asking a specific, awkward question about a known hard process—half rim groove depth is mine—and listening carefully to how fast they answer.

If the answer is vague, that's not a red flag. That's a boundary you haven't hit yet. Find out where it is before the container ships.