Labels used in this module: [Sourced] = checked against a source listed at the end. [General] = standard industry knowledge, not tied to one source.
1. The story: a pack of gum
On the morning of June 26, 1974, a cashier at a Marsh supermarket in Troy, Ohio, picked up a pack of Wrigley's Juicy Fruit and ran it over a laser scanner. It beeped. That was the first grocery item ever scanned with a Universal Product Code. [Sourced: HISTORY, WOSU]
Nobody cheered about the gum. They cheered about what came next: a store could finally know, item by item and in real time, what it sold.
The idea took a long time to arrive. In 1932 a Harvard business student, Wallace Flint, wrote a thesis on automating the checkout line. It was never practical. In 1948, a Drexel graduate student, Bernard Silver, overheard a grocery executive ask the dean for a way to capture product information automatically. Silver teamed up with Joseph Woodland, who drew four lines in beach sand and realised that wide and narrow lines could carry data like Morse code dots and dashes. Their patent was granted on October 11, 1952, but the technology of the day was too clumsy, and they sold it for $15,000. [Sourced: IBM]
What changed by the 1970s? Supermarkets were booming, grocers and manufacturers asked together for one standard symbol, and cheap lasers made scanning practical. IBM engineer George Laurer led the design of the symbol that won. The UPC was introduced in 1973, and the gum was scanned a year later. [Sourced: IBM]
2. The one idea
A store needs two different names for every product: one for the shelf in front of it, and one the whole world agrees on.
- The SKU (stock keeping unit) is your name. A business invents it, for its own use. It tells your systems which exact version of a thing you are counting. [General]
- The UPC is the world's name. It is a number printed as bars that any scanner in any store can read. It is the same in every shop. [Sourced: Wikipedia, GS1]
When someone says "how many do we have?", they are really asking "how many of this one exact SKU, in this one place, right now?"
3. The kitchen-table version
Think of a family with a shared pantry.
You write "Mom's tomato sauce, big jar" on a sticky note. That label means something only in your house. It is your SKU.
Now the jar of Rao's sauce at the store has a printed barcode. Every store in the country can read it. That is the UPC.
Your sticky note can be as detailed as you like. The barcode is fixed by someone else, and every shop must agree to it. You use both: the sticky note to run your house, the barcode to talk to everyone outside it.
4. What exactly is a SKU?
A SKU is the smallest thing you need to count separately. If a customer would be upset to get the wrong one, it is a different SKU. [General]
Take one T-shirt design:
| What you see | How many SKUs? |
|---|---|
| One style, "Classic Tee" | 1 style |
| In 3 colours | 3 colour versions |
| Each in 5 sizes | 15 SKUs (3 x 5) |
The shirt is one product to the customer's eye, but 15 separate things to the warehouse. Run out of "Blue, Medium" and you have not run out of Classic Tee, but you have lost a sale. [General]
Where does the SKU come from? Honest answer: nobody knows who coined it. The abbreviation shows up in print by the mid 1970s, as the shortened form of "stock keeping unit", and it grew out of inventory practice. We found no named inventor. [Sourced: Etymonline says "by 1974"; inventor not found]
Who owns it? You do. Two stores selling the same jar of sauce will each have their own SKU for it. That is why SKUs do not match between companies. [General]
5. What exactly is a UPC?
A UPC-A has 12 digits. [Sourced]
| Part | Digits | What it does |
|---|---|---|
| Number system | 1 | The type of item |
| Manufacturer ID | 5 | Which company made it |
| Product ID | 5 | Which product from that company |
| Check digit | 1 | Catches typing and scanning mistakes |
Outside North America, a 13-digit cousin called EAN-13 is common. The umbrella name for all of these is GTIN (Global Trade Item Number), which comes in 8, 12, 13 or 14 digits. [Sourced: Wikipedia]
The company that issues the numbers is GS1. A maker gets a company prefix from GS1 and then numbers its own products. [General; GS1 is the standards body behind UPC/GTIN]
SKU vs UPC at a glance
| SKU | UPC / GTIN | |
|---|---|---|
| Who makes it up | You | A global standard (GS1) |
| Same across stores? | No | Yes |
| Can carry your own meaning (colour, size, season)? | Yes | No, it is just an ID |
| Used for | Planning, stock, reordering | Scanning, selling, sharing data between companies |
5b. Nuance box: same word, different meaning
The identifier family. Several codes sound alike and do different jobs. [Sourced: industry identifier guides and Google Merchant help; vendor guides, definitions only]
| Code | Whose is it? | What it identifies |
|---|---|---|
| SKU | The seller's own | One sellable variant in that seller's stock system (style, colour, size) |
| GTIN (UPC, EAN, JAN) | GS1 system, assigned by the brand owner | One exact trade item, same everywhere |
| MPN | The manufacturer | The manufacturer's own part number |
| Marketplace ID (for example Amazon's ASIN) | The marketplace | A listing inside that marketplace |
| One code cannot do all five jobs. One product can carry several SKUs at different sellers but should have one GTIN. [General] |
UPC, EAN, JAN. GS1 US describes the EAN as a 13-digit GTIN used across much of the world, and the UPC as a 12-digit code used primarily in North America. [Sourced: GS1 US] The Japanese version, JAN, sits in the same family. [Sourced: barcode symbology guides, secondary] A UPC-A can be written as an EAN-13 by adding a leading zero, which does not change the check digit. [General; Sourced only from a secondary symbology guide]
When does a product get a new GTIN? GS1's GTIN Management Standard exists so that industry makes consistent decisions about the unique identification of trade items, and it distinguishes a new product from a change to an existing one. A new GTIN is needed for a new product, and for changes such as formulation or functionality that affect legally required declared information on the package, among other rules. [Sourced: GS1 GTIN Management Standard pages; I read the overview and one rule, not the full rule list] A common mistake is reusing an old barcode on a changed product. [General; Our view that this is the common mistake] Do not teach the exact list of triggers until we read the whole standard.
6. Worked example: how the check digit catches mistakes
The last digit of a UPC is not random. It is calculated from the other eleven. Let us do it with the example from Wikipedia: 03600029145 + ? [Sourced: Wikipedia]
- Add the digits in odd positions (1st, 3rd, 5th, 7th, 9th, 11th): 0 + 6 + 0 + 2 + 1 + 5 = 14
- Multiply by 3: 14 x 3 = 42
- Add the digits in even positions (2nd, 4th, 6th, 8th, 10th): 3 + 0 + 0 + 9 + 4 = 16
- Add the two: 42 + 16 = 58
- The check digit is whatever brings you up to the next multiple of 10. 58 needs 2 to reach 60. Check digit = 2.
Full code: 036000291452. If a scanner misreads one digit, the sum no longer lands on a multiple of 10 and the register rejects the scan. (We checked this arithmetic ourselves, and it matches the published check digit.)
7. How it works inside a store (today)
[General]
- A shopper brings an item to the register.
- The scanner reads the UPC.
- The store's system turns that UPC into its own SKU and looks up price and description.
- The sale is recorded against the SKU, at that store, at that time.
- Overnight, sales roll up. When stock of a SKU drops below its safety level, a reorder goes to the supplier.
That one scan feeds forecasting, ordering, merchandising and accounting. The rest of this course is about what happens at each of those steps.
8. What changes in an AI world
[Mostly General, with one sourced fact]
- Barcodes are getting bigger. The industry is moving from the classic 1D UPC to 2D codes like QR, which can carry much more than a product ID: batch, expiry date, even a web link. GS1's initiative to get there is called Sunrise 2027 (end of 2027 target for retail point of sale). [Sourced: GS1 US, GS1 Canada]
- Why AI cares. AI is only as good as the labels it learns from. A messy SKU list (same product, three names) gives bad forecasts. Clean product data is now a competitive edge.
- Matching products across the internet. AI tools can read a title and photo and guess which product it is. Useful, but mistakes happen, and an exact ID like a GTIN is still the reliable anchor. [General, our view]
- Shopping agents. When an AI shops for a person, it needs an unambiguous way to say "this exact item". That is a GTIN job. [General, our view; fits the Commerce Stories thesis]
9. Try it yourself (no code, 10 minutes)
- Pick any packaged product at home. Find its barcode.
- Write down the 12 or 13 digits. Split them into manufacturer part and product part.
- Do the check-digit sum on the first 11 digits. Does it match the last digit?
- Practice problem: what is the check digit for 04210000526? (Answer at the bottom.)
- Now invent your own SKU scheme for 3 products you own, such as
SHIRT-BLU-M. Notice how you chose to carry meaning in the code. A UPC cannot do that.
10. Self-check
1. Who creates a SKU? a) GS1 b) The business that sells the item c) The government Answer: b.
2. A shirt comes in 4 colours and 6 sizes. How many SKUs? a) 10 b) 24 c) 6 Answer: b. 4 x 6.
3. What is the last digit of a UPC for? a) Marks the country b) Counts the stock c) Catches reading mistakes Answer: c.
4. Two stores sell the same jar of sauce. Which matches between them? a) The SKU b) The UPC Answer: b. The SKU is each store's own.
5. First item ever scanned? a) A can of soup b) A pack of gum c) A box of cereal Answer: b. Wrigley's Juicy Fruit, Troy, Ohio, June 26, 1974.
Practice answer, step 4: 4 (so the full UPC is 042100005264).
Sources
- IBM, The UPC: https://www.ibm.com/history/upc (Flint 1932, Silver and Woodland 1948, patent Oct 11 1952, $15,000, Laurer, 1973)
- HISTORY, first scan: https://www.history.com/this-day-in-history/june-26/barcode-scanner-invention-first-use
- WOSU on the 50th anniversary: https://www.wosu.org/2024-06-25/50-years-ago-the-first-retail-barcode-was-scanned-at-an-ohio-grocery-store
- Wikipedia, UPC structure and check digit: https://en.wikipedia.org/wiki/Universal_Product_Code
- Etymonline, SKU ("by 1974"): https://www.etymonline.com/word/sku
- GS1 US, Sunrise 2027: https://www.gs1us.org/industries-and-insights/by-topic/sunrise-2027
- GS1 Canada retail 2D FAQ: https://gs1ca.org/gs1ca-components/documents/Retail_2D_FAQ_EN.pdf