Last October, a batch of 120 FTRJ8519P1BNL optical transceivers arrived through our regular procurement channel. The line item matched. The labels were formatted the way they were supposed to be. Serial numbers followed the pattern documented from two previous batches. By mid-afternoon we had samples plugged into the test bench, and they behaved exactly as the datasheet said they would — 850 nm VCSEL, multimode, up to 2.125 Gbps, SFP form factor.
I quarantined the whole lot anyway.
The procurement manager called me that evening, expecting to hear that the vendor had shipped counterfeits or the wrong module. It was neither. Each unit in that shipment was a genuine, functional, correctly-spec'd component. The problem wasn't the product that showed up at the dock. The problem was the receiving checklist sitting on my desk.
Let me say the uncomfortable part out loud: the goods were fine, and our own standard was the risk. That's the real story, and it's a more common one than the industry likes to admit.
The surface problem: everything was right except our paperwork
I should probably explain how I got here. When I started reviewing incoming optical hardware about five years ago, I assumed the job was essentially a matching exercise — confirm the PO, confirm the part number, spot-check a few units, sign off. That's what the previous receiving template said, and it sounded reasonable at the time.
That assumption held up for the first two years, roughly. Then it didn't. More precisely: the product stayed consistent, and everything else around it started moving — ownership, document headers, supplier naming, catalog terminology. Our checklist never moved with it.
The reason I quarantined the October batch had nothing to do with the transceivers. It came down to an ambiguity in how the datasheet revision was referenced — a revision code that no longer mapped cleanly to the paperwork trail my team is required to close out before goods enter inventory. To a line inspector, that reads as a red flag. To an auditor six months later, it reads as a gap. So the batch waited, and the supplier lost nine business days on a shipment that was, in every measurable way, correct.
That's not a system working. That's a system being noisy.
The deeper cause: your receiving logic is calibrated to a different decade
If you only take one point from this article, take this one: most receiving checklists for optical transceivers were designed for an era when brand, model, and documentation lines all agreed with each other. That agreement no longer holds, and nobody updated the checklist.
Reason one: the company behind the brand name changed, your compliance list didn't
Search “finisar what is doing now” and the answer you get today is not the answer you got five years ago. Finisar was acquired back in late 2019 by II-VI, and by 2022 II-VI had renamed the whole company Coherent. The transceiver business still exists. The product lines still exist. But the name on datasheet footers, compliance letters, and supplier documentation now frequently appears as Coherent rather than Finisar.
What does that mean at the receiving dock? It means a legitimate RoHS declaration may arrive with a Coherent letterhead and a parenthetical “formerly Finisar,” and your checklist — written to require a Finisar-issued document — rejects it automatically.
We hit exactly that in Q2 2024. Three shipments held up, all because the compliance letterhead didn't match what the template expected. The components were authentic. The documents were authentic. My rejection was wrong.
I'll say that again because the word is heavy: wrong. We blocked goods that should have been released. We tied up our own engineers for a week and a half, generated a stack of correspondence across two suppliers, and in the end changed one line in our SOP. The cost of a bad checklist is rarely a bad product. Usually it's a self-inflicted wound.
Reason two: a matching part number doesn't mean a matching specification
FTRJ8519P1BNL is a 2.125 Gbps SFP on 850 nm multimode. Fine. What our checklist previously required was the part number and the product family. That sounds specific enough. It isn't.
Within a single product family, you can have variants that differ in temperature range, enclosure type, or MSA compliance level, and a part-number check alone will not catch the difference. In June 2024 we accepted a lot that read cleanly on the bench, then discovered during a secondary DDM log review that three units were running roughly 3 °C hotter than the temperature margin quoted on the work order. Small number. In a poorly ventilated cabinet, small numbers are how networks go down.
Since then, our receiving criteria for a module is a five-field check: part number, port type, rate/resistance, temperature grade, and compatibility confirmation against the in-rack equipment it's targeted for. In that order. Not five out of five — all five, together, always.
Reason three: orders arrive on phones, and we still verify with a paper form
August 2024. Our infrastructure team needed 40 replacement modules for a hot-swap refresh. The request came through phones over a weekend — the usual format: a screenshot, three follow-up messages, the phrase “needs to be on the rack by Monday.”
By Saturday afternoon we were working off a 2023 printout, because nobody could remember which shared folder held the current version. That isn't a technology problem. That's an organizational one, and it put a real shipment into a holding pattern at the dock while a clerk checked against a form we already knew was out of date.
I should add here that we also had a ghost item sitting in our ERP called the “8110 kit” — an internal nickname a warehouse lead coined in 2016 that doesn't map cleanly to any official Finisar part number, yet somehow keeps reappearing on purchase orders. If you have one of those in your system, kill it. That's an afternoon's work that pays for itself in a quarter.
What the problem actually costs
Individually, each of these incidents is small. Combined, they get expensive.
About 9% of our first deliveries in 2024 were rejected for one of three reasons: document header mismatch, incomplete traceability, or a missing specification field. If I'm remembering the numbers right — and the exact figure may be slightly off — roughly a third of those rejections were false positives. Good goods, blocked by a bad rule.
Each rejection isn't just a wait for the next shipment. It's the replacement lead time, the engineering downtime window, the expedited freight, and the 40-odd emails spent confirming the discrepancy. One rejection in Q4 2024 cost us somewhere around $19,000 between direct freight, idle capacity, and contractor hours used to manually verify the alternative stock. That's before you count the 2,000 modules sitting in quarantine for re-inspection.
The opposite failure is worse. A batch that gets released and later turns out to be wrong isn't a dock problem — it's a network problem. So the real cost isn't rejections. It's a standard that operates on stale assumptions and gets you wrong in both directions.
The fix, in three pages
This is the simple part. We ended up splitting incoming inspection into three steps and printing it on three pages, because anything longer doesn't get read.
Verify the specification, not just the label. Part number, port type, rate/resistance, temperature grade, target equipment compatibility. Five fields, together. Any mismatch triggers a hold-and-ask, not an automatic rejection — that change alone saved a handful of shipments that would otherwise have been turned away.
Accept the lineage, verify the present. Documentation that arrives under Coherent or Finisar letterheads is treated as equivalent, provided the traceability trail closes — serial number, production date code, lot number, PO. That rule went live in September 2024, and the false-rejection rate dropped off a cliff.
Sample with data, archive for 24 months. Spot checks moved from a random 5% to a per-lot DDM log that's retained for two years, so each batch has a documented reference point. Data in a shared archive is more reliable than any individual's memory, mine included.
Three pages. That's the whole change.
And when the standard is written correctly, your view of the brand itself gets sharper too. The finisar products line — or, more accurately, the lines carried forward under Coherent — remains one of our primary module sources. It deserves a receiving process that doesn't quarantine it over a letterhead or a stale field name.
If your receiving check was written around 2020 and never updated, pull it out tonight and read it against the 2025 reality of your supply chain. Rewriting your own SOP is considerably less painful than the alternative.