Daniel sent us this one — and I suspect half our listeners just felt a sympathetic twitch in their eye. You're staring at a spec sheet for a power supply. The manufacturer's site calls it Model ABC dash one two three. Digi-Key has it as MPN colon ABC dash one two three dash B. Amazon says SKU colon nine eight seven six five. And the invoice you actually received says Part Number colon one two three dash four five six seven. You need to order ten more. Which of those four codes do you actually type into the search bar?
If you pick the wrong one, you either get nothing, or you get the right-looking box with the wrong internals, and now you've got a hundred and ten volt unit on a two-twenty volt rail and a very bad afternoon.
What's actually going on here? Let's start by defining the players in this identifier ecosystem.
The core problem is that product identifiers are not standardized across the supply chain. A single physical widget can easily have five or more identifiers, and each one exists because a different organization needed to label it for a different purpose. You've got the manufacturer who created the thing, the distributor who resells it, the retailer who sells to the end user, and the standards bodies like GS1 who issue universal codes. Every layer adds or transforms identifiers. And none of them talk to each other in a way that makes your life easier.
It's the Tower of Babel, but everyone's selling voltage regulators.
So if you're a technical buyer — an engineer, a procurement person, someone repairing equipment — you care about this because using the wrong identifier doesn't just mean you get the wrong thing. It means you lose weeks of lead time, you eat return shipping costs, and in the worst case you install a counterfeit or incompatible part and blow something up. With global supply chain fragmentation and counterfeit parts on the rise, knowing the canonical identifier is a survival skill.
The counterfeit angle is real. If you search by a fuzzy model number on a marketplace, you're basically inviting the algorithm to serve up whatever looks close. Some of those listings are genuine alternates. Some are e-waste with a fresh sticker.
Let's lay out the taxonomy quickly before we go deep. The key players in any spec sheet are the MPN, or Manufacturer Part Number — that's the canonical identifier assigned by the company that made the thing. The model number, which is usually a marketing-facing name for a product family. The SKU, or Stock Keeping Unit, which is an internal inventory code used by whoever's selling it to you. The GTIN, which is the umbrella term for UPCs in North America and EANs in Europe — those are the barcode numbers. And then you've got distributor-specific catalog numbers, OEM part numbers, and vendor reference codes. Each one answers a different question, and none of them is the whole story.
With that framework in mind, let's tackle the most common confusion first: part numbers versus model numbers.
This is the one that trips people up constantly. A model number is a marketing-facing identifier for a product family. Think Intel Core i seven dash thirteen seven hundred K. That tells you it's a thirteenth-gen chip, i seven tier, unlocked. But it doesn't tell you which specific stepping of the silicon, which packaging option, or whether it's tray or boxed. A part number — or MPN — is a manufacturing-facing identifier for a specific SKU, often including revision, packaging, temperature range, or region codes.
The model number is the family name. The MPN is the full legal name including middle initial and the part where your parents are mad at you.
And they overlap when the product is simple enough that there's only one version. A basic through-hole resistor might have the same string for both model number and MPN because there's nothing to disambiguate. But the moment you've got multiple package types, voltage ratings, or temperature grades, the model number stays the same and the MPN sprouts suffixes.
Let's make this concrete. You mentioned the Texas Instruments LM three five eight.
The LM three five eight is a dual operational amplifier. The model number is LM three five eight. That's what you search when you want to know what the part does. But the MPN might be LM three five eight N for the DIP package, or LM three five eight D for the SOIC surface-mount package. Those are different physical parts. Same silicon die inside, but different form factors, different pin spacings, different thermal characteristics. If you order by model number alone, you're flipping a coin.
Then Digi-Key assigns it a completely different number.
Digi-Key's internal SKU for that LM three five eight N might be two nine six dash one two three four five dash ND. Mouser's SKU for the exact same physical part might be five nine five dash LM three five eight N. Amazon's ASIN is yet another string. And the UPC on the bag — if there is one — often maps to the product family, not the specific revision.
The same physical bag of op-amps now has five identifiers, and only one of them — the MPN — is guaranteed to work across every distributor's search bar.
That's the practical rule. For comparing quotes or finding replacements, always search by the MPN. It's the only identifier that's consistent across the supply chain. Distributor SKUs are useless for cross-referencing because they're internal — Digi-Key's SKU means nothing to Mouser. Model numbers are too broad — they'll return every package variant and sometimes unrelated parts that happen to share a prefix.
I want to underline something you just said, because it's the kind of thing people learn the hard way. The UPC, the barcode, the GTIN — it's not a product identifier. It's a trade item identifier. GS1, the organization behind GTINs, was founded in nineteen seventy-three and has issued over one billion unique GTINs as of two thousand twenty-five. But a GTIN answers the question "what is this thing on the shelf at the point of sale," not "what is the exact revision of this component.
That distinction matters enormously. Two revisions of the same product can share a UPC. The manufacturer considers them close enough for retail purposes. But if revision A has a different input voltage range than revision B, and your application needs revision B specifically, the UPC will not save you. You'll scan the barcode, the system says "yes, this is the product family you want," and you get revision A in the mail.
Which brings us to obsolete parts. You're repairing a piece of equipment from the nineties, you pull up the board, it says seventy-four LS zero zero N. That part is long out of production. What do you search?
You search the MPN — seventy-four LS zero zero N — and you look for a cross-reference or supersession table. Most major manufacturers publish these. They'll say "the seventy-four LS zero zero N is obsolete; use the seventy-four HCT zero zero N as a drop-in replacement." If you had searched by model number — just seventy-four LS zero zero — you'd get a flood of results: different logic families, different packages, parts that are vaguely related but not what you need. The MPN pinpoints the exact package and specification, and the cross-reference table gives you the modern equivalent. Distributor SKUs won't help you here at all — they're tied to inventory that no longer exists.
There's something almost poignant about a Digi-Key SKU for an obsolete part. It's a tombstone in a database.
A little digital fossil. Now, let's talk about why this identifier chaos exists in the first place, because it's not an accident. Manufacturers assign MPNs for their own internal use — engineering, production planning, quality control. The MPN encodes information the manufacturer needs: which factory line, which revision, which test program. Distributors like Digi-Key and Mouser receive parts from thousands of manufacturers, and they need their own inventory management systems. So they assign internal SKUs that fit their warehouse software, their picking systems, their order fulfillment workflows. Amazon assigns an ASIN because everything in Amazon's catalog needs an Amazon Standard Identification Number — it's the database key that makes their entire system work.
Retailers layer on vendor catalog numbers because their ERP system was set up in nineteen ninety-seven and nobody wants to touch it.
I've seen systems where internal product codes were derived from supplier part numbers, and then the supplier changed their numbering scheme, and suddenly every downstream database had phantom references that didn't match anything real. It was like a game of telephone where the first player walked away.
The chaos isn't a bug. It's a stack of legacy decisions, each one rational at the time, that collectively produce a spec sheet that looks like it was typed by a cat.
Now we get to the second big question. Let's say you've got the spec sheet in front of you. How do you actually identify the canonical identifier — the one you should be searching?
First rule: look for the field labeled MPN or Manufacturer Part Number or Mfr period Part Number sign. That's the one. If the sheet only says Part Number or Model Number without qualification, you need to go upstream. Download the manufacturer's official datasheet — not the distributor's summary page, not the SEO-spam blog that regurgitated the specs — and find the table labeled Ordering Information.
Every serious component manufacturer includes an Ordering Information table in their datasheet. It's usually near the end. It lists every valid MPN for that product family, with columns explaining what each suffix means. This table is the source of truth. If the identifier on the spec sheet you're looking at doesn't appear in that table, it's not the MPN — it's someone else's internal code.
How do you tell when a vendor is showing you their own internal SKU versus the manufacturer's part number?
Internal SKUs usually have prefixes or suffixes that match the distributor's naming convention. Digi-Key SKUs follow a six-to-eight-digit numeric code with a two-letter suffix — that ND at the end stands for Newark Digi-Key, a legacy from their merger. Mouser numbers often start with a three-digit manufacturer code. If you see an identifier with dashes and letters that don't match the manufacturer's datasheet, it's almost certainly an internal SKU. Cross-reference it with the manufacturer's website. If the manufacturer's site doesn't recognize it, it's not their number.
Now that we know how to distinguish MPNs from model numbers, let's look at when those model numbers actually tell you something useful — and when they're just opaque codes.
This is where it gets genuinely interesting. Some model numbers are structured — they encode meaning that a human can decode. Intel's consumer CPU naming is the gold standard here. Core i seven dash thirteen seven hundred K. The i seven is the tier. The thirteen means thirteenth generation. The seven hundred is the specific SKU within that tier. The K suffix means unlocked multiplier — you can overclock it. An F would mean no integrated graphics. A T would mean power-optimized. This scheme was introduced in two thousand eight with the Core i seven nine hundred series, replacing the older Pentium numbering, and it's remarkably consistent. A knowledgeable buyer can look at an Intel model number and immediately know the generation, performance tier, and feature set.
It's the IKEA naming system, but for silicon. You know what a Billy bookcase is, and you know what a Kallax is, and you never confuse them.
But contrast that with an industrial component. Take a Murata capacitor. The part number is something like GRM one five five R seven one H one zero four K E one four D. That is not designed for human parsing. It encodes size — zero four zero two — dielectric type, capacitance, voltage rating, and tolerance. But you can't read it unless you have Murata's decoding scheme in front of you. It's designed for database lookups, not for a human to glance at and understand.
The Murata code is basically a compressed file. The Intel code is a headline.
That's a perfect way to put it. And the key insight for a buyer is to recognize which kind you're dealing with. A structured model number tells you something at a glance — you can compare two Intel chips by reading the names. An opaque registry ID tells you nothing without the datasheet. Siemens PLC model numbers like six E S seven two one one dash one B E four zero dash zero X B zero are in this category. They look like someone dropped a Scrabble set, but they're precise database keys. You search them, you don't interpret them.
Which means the buyer's workflow depends on the kind of number they're holding. Structured number: you can reason about it. Opaque number: you verify it against the Ordering Information table and move on.
And now we need to talk about the gotcha that burns people the most: revision suffixes.
The dash A, dash B, dash REV two situation.
These look cosmetic. They're a couple of characters at the end of a long string. But they can indicate non-interchangeable parts with different specifications, pinouts, voltage ratings, or firmware. The classic example is the Raspberry Pi four Model B. Revision one point one and revision one point two. The model number is the same — Raspberry Pi four Model B. But revision one point two, released in twenty twenty, changed the USB-C power controller. Some chargers that worked fine on revision one point one would not power revision one point two. Same model number. Same product family.
If you were ordering by model number alone, you'd have no idea which one was arriving.
Another example: Mean Well power supplies. The LRS dash three fifty dash twenty four is a popular industrial power supply. Revision A and Revision B have different input voltage ranges. Order the wrong revision for your region, and you've got a unit that expects a hundred and ten volts on a two-twenty volt supply — or vice versa. The model number on the box looks identical. Only the suffix changes.
The rule is: treat revision suffixes as potentially non-interchangeable unless the manufacturer explicitly states backward compatibility in the revision notes or errata document. And actually read those notes. Don't assume.
The suffix is cosmetic only if the manufacturer says it is. If they're silent, assume the revisions differ. Most reputable manufacturers publish a Product Change Notification, or PCN, when a revision changes something meaningful. Find that document. It'll tell you whether it's a minor tweak or a functional change.
I want to pull on a thread you mentioned earlier — the structured versus opaque distinction — because it connects to something bigger. The trend over the past twenty years has been away from human-readable model numbers and toward opaque registry IDs. Intel's scheme is a holdout. Most new components, especially in industrial and automotive, use strings that only make sense to a computer.
There's a reason for that. As product lines proliferate — more variants, more regional SKUs, more regulatory requirements — the number of parameters you need to encode exceeds what a human-readable scheme can handle. Intel can get away with a structured name because a CPU has a relatively small set of differentiating features: cores, clock speed, graphics, unlock status. A capacitor has dozens: capacitance, tolerance, voltage, dielectric, temperature coefficient, size, termination style, packaging, and on. You can't fit all that into a name a human wants to read.
The opaque code isn't user-hostile. It's just the compression format you need when the data gets dense enough.
And this is where the buyer's skill comes in. You don't need to decode the Murata part number. You need to know how to look it up. The datasheet is the decoder ring. The Ordering Information table is the Rosetta Stone.
Let's talk about a specific workflow. You're staring at a spec sheet. It has five fields that all look like identifiers. What's your move?
Step one: find the field labeled MPN, Mfr period Part Number, or Manufacturer Part Number. That's your primary key. Step two: if that field doesn't exist, look for Part Number or Model Number, then go to the manufacturer's datasheet and find the Ordering Information table. Match what you see on the spec sheet to what's in that table. Step three: if the identifier on the spec sheet doesn't appear anywhere in the manufacturer's datasheet, it's a distributor SKU or vendor code. Search it on the distributor's site to confirm, then extract the real MPN from their product page. Step four: once you have the MPN, use that for everything — quote comparison, replacement search, inventory check.
If you're dealing with an obsolete part, the MPN is your only reliable entry point into cross-reference databases.
Sites like Octopart, FindChips, and the distributor search tools all index by MPN. If you feed them a distributor SKU, they won't know what to do with it. If you feed them a model number, you'll get noise. The MPN is the skeleton key.
There's a whole subculture of procurement engineers who maintain personal databases of MPNs for exactly this reason. They've been burned before, and they're not getting burned again.
That's good practice. If you're ordering the same parts repeatedly, keep a spreadsheet with the MPN, the revision you qualified, and the date you last verified it. When the manufacturer obsoletes that revision, you'll know to check the PCN and qualify the replacement before your production line stops.
I want to loop back to something about GTINs, because there's a misconception that a UPC uniquely identifies a product. It doesn't. It identifies a packaged trade item at the point of sale. GS1 will happily issue the same GTIN for two revisions of a product if the manufacturer considers them retail-equivalent. Your inventory system might care deeply about the revision. The barcode does not.
This is why the GS1 GTIN lookup tool — at gepir dot gs one dot org — is useful but limited. It'll tell you what company owns that barcode prefix and what product family it's assigned to. It won't tell you the revision. For that, you need the MPN.
To summarize the identifier hierarchy: MPN at the top. That's the canonical key. Then model number — useful for understanding the product family, but not precise enough for ordering. Then GTIN or UPC — good for retail scanning, useless for revision-level identification. Then distributor SKUs and ASINs — useful only within that distributor's ecosystem. You can't cross-reference them, you can't search datasheets with them, and they'll stop working the moment the distributor changes their system.
That hierarchy has real consequences. I've seen procurement teams order by model number, receive the wrong package variant, and spend six weeks reordering. I've seen repair shops search by a distributor SKU from an old invoice, find nothing, and assume the part is unobtainable when it's actually in stock everywhere under its MPN. These aren't theoretical problems.
The distributor SKU graveyard is real. Digi-Key changed their numbering system at some point in their history, and old SKUs just... If you'd been ordering by Digi-Key SKU in your ERP, you suddenly had no way to reorder.
Which is why the MPN is the only identifier you should store in your own systems. Everything else is ephemeral.
Let's talk about one more gotcha: regional variants. Same model number, different MPNs for different regions, because the power supply or the wireless certification or the packaging is different.
This is especially common in consumer electronics and power supplies. A model number like LRS dash three fifty dash twenty four might have different MPNs for the North American version, the European version, and the Chinese version. The model number is identical. The MPNs differ. If you're in Europe and you accidentally order the North American MPN, you get a unit with the wrong input voltage and the wrong plug. The model number won't warn you.
This is where the Ordering Information table in the datasheet becomes essential. It'll list every regional variant with its specific MPN and the region it's certified for.
Always check that table. It's the difference between receiving a part you can use and receiving a paperweight.
All of this complexity leads to a few concrete rules you can apply the next time you're staring at a spec sheet.
Rule one: always search by MPN when comparing quotes or looking for replacements. Ignore model numbers, SKUs, and ASINs for cross-referencing. They'll lead you astray.
Rule two: when a spec sheet shows multiple identifiers, look for the field labeled MPN or Mfr period Part Number. If it's ambiguous, download the manufacturer's official datasheet and find the Ordering Information table. That table is the source of truth. If the identifier you're holding isn't in that table, it's not the manufacturer's number.
Rule three: treat revision suffixes — dash A, dash B, dash REV two — as potentially non-interchangeable unless the manufacturer explicitly states backward compatibility. Check the revision history. Check the errata document. Check the PCN. Five minutes of reading can save you weeks of headache.
An additional practical step: bookmark the GS1 GTIN lookup tool at gepir dot gs one dot org. It's not going to solve your revision problems, but it'll help you verify whether a barcode actually belongs to the manufacturer you think it does. Counterfeit parts often carry cloned or mismatched barcodes.
If you're ordering frequently, build a personal database of MPNs for your commonly used parts, including the specific revision you've qualified. When the manufacturer sends a PCN about an upcoming change, you'll know exactly which of your products are affected.
There's a broader question here, and I think it's worth asking: where is this all heading? We've got AI-generated spec sheets flooding search results, SEO-spam product pages that copy-paste specs from three different revisions, and marketplaces that prioritize fast shipping over accurate identification. Is the identifier problem going to get better or worse?
I think short-term, it gets worse. The trend on Amazon and other marketplaces is toward private-label SKUs and consolidated listings where multiple sellers offer "the same" product under different internal codes. The listing might show a model number that's actually just the seller's inventory code. The photo might be of revision A while they're shipping revision C. The AI-generated description might mix specs from incompatible variants. The burden of verification is shifting entirely onto the buyer.
The Amazon ASIN is becoming the default identifier for a generation of buyers, and the ASIN tells you nothing except "this is a thing someone is selling on Amazon.
It only works inside Amazon's ecosystem. You can't take an ASIN to Digi-Key. You can't search a datasheet by ASIN. It's a walled-garden identifier that traps you inside one retailer's catalog.
On the other hand, the rise of digital twins and Product Lifecycle Management systems might eventually force some standardization. If every physical part has a digital twin that needs to be tracked across its entire lifecycle, you need a persistent, universal identifier. The MPN is the natural candidate.
GS1 is pushing in this direction with their Digital Link standard, which embeds GTINs in web-readable formats. But GTINs still don't solve the revision problem. Until the standards bodies and the manufacturers agree on a revision-level identifier that's universally adopted — and I'm not holding my breath — the MPN remains the best tool we've got.
The takeaway for anyone listening is: the alphabet soup isn't going away. But you can navigate it. MPN is your compass. The Ordering Information table is your map. And the revision suffix is the fine print you ignore at your peril.
Next time you're staring at a spec sheet, ask yourself: is this the MPN, or is it just a model number? If you can't answer that question confidently, you haven't finished reading yet.
Now: Hilbert's daily fun fact.
Hilbert: In the eighteen-tens, in the colony of Demerara in what is now Guyana, there lived a man named Cuffy who worked as a professional "cricket catcher" — his sole job was to capture singing crickets and deliver them alive to wealthy plantation owners who believed the sound warded off malaria-carrying mosquitoes. He was paid per cricket, and local records suggest he developed a symbiotic working relationship with a troop of capuchin monkeys who would locate the crickets in exchange for the fruit he carried in his pockets.
A cricket-based mosquito defense system with monkey subcontractors. I respect the hustle.
The interspecies supply chain.
This has been My Weird Prompts. Thank you to our producer, Hilbert Flumingtop. You can find every episode at myweirdprompts dot com or on Spotify. If you've got a spec sheet horror story, we want to hear it — drop us a line.
