← the whole review · The mechanism · The twelve shelves · The names · The labels not on your list · Candidates from your old lists · Brand & Identity · The open calls

Retail & Commerce

What counts as a retail piece, and the display pick

R-01The display pick · full

Is a store display named by the specific thing it is — an endcap, a point-of-purchase unit, a window, an aisle — or by one general word, with the part of the store recorded separately?

Start from what you already have. Every piece in the library carries one word that says what kind of thing it is — a brand film, a press release, a poster; that word is its label. The labels themselves are browsed in groups, and a shelf is simply one of those groups. Your approved list has one in-store label on it, Retail Display Piece, and then your own expansion table went further and named four sharper ones: endcap display, point-of-purchase display, window display and aisle display. Keeping all five would mean two ways to say the same thing, so this is the pick.

The argument for the four specific names is the test you have applied to every other row, and it is worth naming here because most of this page turns on it: the kind test. Does the name describe a recognisable kind of thing, something a person would identify on sight — rather than a fact about where the piece happened to run? A merchandiser recognises an endcap the way an editor recognises a press release: it is bought differently, built differently and photographed differently. On that reading the general word is the weak one, because “retail display piece” is what you write when you have not looked.

Here is the evidence that cuts the other way, which is why this is on the page as its own decision rather than folded into a batch. It is worth being precise about whose standard this is: the published standard for advertising inside shops was written by the advertising industry's own standards body, not by the retail trade. It splits a store into five places a piece can be — outside, at the entrance, at the checkout, in the aisle, and on connected screens — and then keeps the names of the pieces themselves small and physical: shelf talker, hanging sign, floor graphic, screen. Read that way an endcap is not a kind of thing at all; it is a spot on the shop floor, and what stands in it is a printed sign or a screen. That shape is closer to option (b): one general display word, plus a separate answer for where in the store the piece stood.

The library cannot settle it for you, because it holds no store display at all — retail collection has never run. The nearest real things it does hold come from one OpenAI post that imagines the launch campaign for the wheel. One image is a paper sheet pinned to a cork board, headed “INTRODUCING THE WHEEL”; another is a horse-drawn carriage with “Stop trolling — start rolling!” painted down its side. Both argue for naming the thing rather than the spot: the sheet is a poster whether it is pinned in a window, at a till or on a wall, and the carriage is a branded vehicle down whichever street it is driven. Neither is stored as anything of the sort today — both say social post, which is where they ran, not what they are.

One thing is worth knowing before you choose, because it is easy to assume the opposite: the stored list of words does already have a group with retail in its name. It is not yours and it is not about shops. It is one of the ten family names that came with the database rather than from you — “Event / Retail / Interactive” — and the 7 words filed under it are event coverage, livestream, event photo, virtual try-on, product configurator, interactive experience and augmented-reality filter. Not one of them is a display, and all 76 of the pieces sitting under that group are event coverage, livestreams or event photos. So whichever option you mark, the words picked here are the first shop-floor words the library will ever have had.

Whichever way you mark it, nothing is written on any piece and nothing moves. One thing to keep consistent: the twelve-shelves page shows Retail & Commerce holding exactly one label, Retail Display Piece, in decision S-10. That is the label option (a) retires, so mark this one and S-10 the same way.

My take: My take is (a), keep the specific names. The kind test is the rule you have used on every other row, and where in the store a piece stood can still be recorded later as its own answer — picking (a) refuses to make that the label, it does not throw it away.

What this call does to everything else written on these pieces

Nothing else on a piece is touched. The one word is a single fact among many, and the rest are written by different work: the pinned-poster image keeps OpenAI as its brand, its three crafts (brand copywriting, graphic design and illustration), the product Codex, the two jobs it does — awareness and community engagement — and its eight themes, whichever way you mark this. The carriage image likewise keeps its two crafts and its six themes. Neither belongs to a campaign, and a still image has no moments inside it, so those stay empty for reasons that have nothing to do with this decision. The only thing this call moves is the word you would browse by — and today not one piece in the library carries any of the words in question.

The alternative, and its cost: (b) keep one general in-store word and record which part of the store the piece stood in as a separate answer. Its cost is two-fold: browsing retail gives you one undifferentiated pile until that separate set of answers is built, and it does not exist yet, so the cost is real work before anything is browsable; and you give up the word a merchandiser would actually type into a search. Its benefit is that it lines up with that same advertising standard, so anything you ever take in from people who work to it needs no translation.

Where this was worked out: the full Retail & Commerce list on the recommendation page, and the advertising industry's in-store standard summarised on the evidence page

R-02Aisle display or endcap · short

Is an aisle display its own kind of piece, or an endcap by another name?

Both are display units standing in a shop aisle, and your table flagged the overlap and then deliberately declined to lean: a merchandiser would keep them apart, because an endcap is premium placement that is bought and built differently, while anyone tidying a word list would fold them into one. Nothing is stored under either word, so this costs nothing whichever way it goes — it is only a question of how you will want to search in a year. The nearest thing the library holds is the pinned paper sheet from the wheel-launch post, the sort of piece that will sit beside both of these the day retail collection runs.

My take: No recommendation here, deliberately — it is genuinely level, and the pick belongs to whoever will be doing the searching. If you have no instinct either way, say so and I will leave it open rather than guess on your behalf.

Where this was worked out: the Retail & Commerce list on the recommendation page, where this row is flagged as overlapping and left open

R-03Interactive kiosk — which group browses it · short

A mall kiosk or an information booth is plainly a kind of piece; which group should browse it — Retail & Commerce, or Interactive & Digital Experience?

Nobody disputes the kind here: a kiosk is a real, recognisable thing, and the only open question is where you would go looking for one — among the shop-floor pieces, because it is a physical commerce fixture, or among the interactive ones, because you walk up and touch it. That choice is never written on a piece, so it can be moved later at no cost; it decides which list the word turns up in when you browse, and nothing else. The nearest thing the library holds is the Anthropic Mars feature page capture — a screen you interact with, filed today by where it ran rather than by the furniture it sits in — which is exactly the ambiguity a kiosk inherits.

My take: No lean from me on this one. Both readings are defensible and it is a taste call about your own browsing, not about what is true of the piece.

Where this was worked out: the Retail & Commerce list on the recommendation page, where the kind passes and only the group is left open

R-04Product detail page · full

Is a product detail page a kind of piece, or the page that pieces run on?

You seeded product detail pages as a kind of piece, and the pushback comes from a ruling you already made rather than from anything new: the smallest thing we tag is the piece itself, and a page is where several pieces are composed together. The biggest commerce platform's own vocabulary says the same thing about the enhanced content brands buy for a product page — it names those modules as pieces placed onto the page, never as the page.

On real rows this is already how it behaves. The Anthropic Mars feature page produced two stored pieces, the whole-page capture and the opening screen below; both are stored as feature articles, and both already answer “product detail” to the question of where they ran, as 20 pieces in the library do today. Nothing about that arrangement is straining, and no one has had to invent a word for the page itself to make it work.

Wanting the whole-page capture to be a kind of its own is a fair thing to want — a full capture is a real object, you can point at it, and it is often the thing a designer wants to look at. But granting it reopens the ruling underneath, and it does not stop at this one row. A marketplace listing and an app-store page are the same shape, so whatever you decide here decides for them; and every piece already cut out of such a page would need a rule for how it relates to the whole. That is why this one is worth marking deliberately rather than nodding through: the cost is not the row, it is the rule.

My take: My take is that the page is a surface, not a kind. If you disagree, tell me what a whole-page capture's word would actually be, and I will take that as the amendment and work out what follows.

What this call does to everything else written on these pieces

Nothing else on either capture is touched, and nothing is rewritten. Both keep Anthropic as their brand and Claude as the product they are about; the whole-page capture keeps its two crafts, event photography and brand copywriting, the five jobs it does and its eleven themes, and the opening screen keeps brand copywriting, its three jobs and its eight themes. Neither belongs to a campaign, and a still capture holds no moments inside it, so those stay empty for reasons that have nothing to do with this call. The only fact this decision touches is where a piece ran — and both captures already carry the answer “product detail” there, so under option (a) not a single row is edited.

The alternative, and its cost: (b) amend the ruling so that a whole-page capture is a kind of piece in its own right. Its cost is larger than the one row looks: you would have to say what that kind is called and what it means for the pieces cut from the same page, and the same amendment then arrives for marketplace listings, app-store pages and every other composed page, each asking again whether the page or the pieces on it is the thing being filed. Its benefit is real too — a whole page is how a designer actually looks at a page, and today the library can only hand you the parts.

Where this was worked out: the product-page pushback under the Retail & Commerce list, and the commerce-platform evidence on the evidence page

R-05Corporate gift · short

Is a branded gift its own kind of piece, or the same kind as merchandise?

A corporate gift is the same physical branded object as merchandise with a different recipient, which is why your table proposed folding it in — and the library makes the case better than the argument does. In the same wheel-launch post there is a hoodie shot on white and a metal keyring hanging from a car key, and no reader would call one merchandise and the other a different kind of thing; what separates them is who paid and who received, which is not what the word on a piece is for. A gifting programme might one day deserve its own view, but a view is a way of looking rather than a kind of thing, and nothing is stored under either word today.

My take: My take is fold. Keep it separate only if you can picture a gift that would look wrong turning up in a merchandise search.

Where this was worked out: the Retail & Commerce list on the recommendation page, where this row is proposed as a fold

R-06The clean retail kinds · short

Do these eleven names join Retail & Commerce as one batch?

Eleven names pass the kind test — each one names a recognisable kind of thing rather than a fact about where a piece ran — and each one traces back to a source of yours: endcap display, point-of-purchase display, window display, retail signage, packaging, unboxing, catalog, brochure, poster, merchandise and branded vehicle, gathered from your seed lines, your touchpoints sheet of 8 July this year and the environment sections of the guideline books. All eleven are new, because retail collection has never run: this group gets its vocabulary before it gets its pipeline, on purpose, because the words are chosen for what is worth capturing rather than for what happens to be in the library this week — and unboxing is deliberately named without its medium: a video of an unboxing and a photo set of the same thing are both filed as unboxing, and the piece itself tells you which one it is — so you would see “unboxing video” on one and “unboxing photos” on the other, without two labels. The library is emptier on paper than in fact, though: the wheel-launch post alone holds a pinned paper poster, a folded printed brochure and a horse-drawn carriage carrying an advertisement, and all 10 of its images are stored as social post — the word for where they ran, not for what they are.

My take: My take is confirm all eleven. If one of them makes you hesitate, strike it in your note — a name that has to be explained twice is the wrong name.

Where this was worked out: the full Retail & Commerce list on the recommendation page, one row per candidate with its verdict

R-07The three not-kinds · short

Three more ideas from your table are real things — but does any of them name a kind of piece?

In-store digital signage is a display whose medium happens to be a screen, so it belongs inside whichever display word survives R-01 rather than beside it; shoppable content describes what a piece can do where it is running, not what it is; and a sales or promotion banner is an advertising format carrying a promotional intent, while the piece itself stays whatever kind it already was. You can watch that behave correctly on two real rows: the opening screen of the Anthropic Mars page is a full-width banner by any designer's reckoning and is stored as a feature article, and the pinned wheel poster is a sales sheet down to its benefits list and its new-and-improved promise, and is stored as a social post — in both cases the promotional intent is a fact about the piece, never the name of it. Confirm the three redirections as one batch and each idea waits for the question that really answers it: where a piece ran, and what it was for.

My take: My take is confirm. None of the three fails because it is unimportant — they fail because they answer a different question, and putting them on the list of kinds would make that question harder to ask later.

Where this was worked out: the fold, surface and placement rows of the Retail & Commerce list on the recommendation page