Appearance
Decided — what has been ruled, and why
Every entry here was once an open question. It was ruled on, the answer was written into the model, and the reasoning was kept so nobody has to reconstruct it later.
Keeping the reasoning is the point. A decision recorded as a bare rule gets re-argued the first time someone finds it inconvenient; a decision recorded with its reasoning can be disagreed with on the merits, which is a much shorter conversation.
Ruled on the first sitting (11 September 2026)
Joe walked the first sitting — five words the library had been writing that were never on his list — and dictated a first pass, then answered four follow-up questions. His words are kept untouched in the banked walk record.
Infographic stays a word, for designed data graphics only.Why (his words): "So, I think infographic matters." The 62 pieces carrying the word came from two places. Many are pages of reports run as ads, and some are research charts taken from web pages. A chart inside an article and a page of a report are parts of something larger, not infographics. How those parts are recorded is a question for the next sitting.
Behind the scenes stays a word.Why (his words): "behind-the-scenes we should definitely keep it's definitely becoming a big big one."
System card stays a word, listed under Documentation & Technical, with a clear line drawn against model card.Why (his words): "System card and model card are definitely important. Let's just make sure it's clear which one is which and how to define it in terms of the prompt and the categorization." And on where it goes: "a documentation in technical groups is a good place to put this." No piece in the library is titled a model card; fifty-five are titled system card, and thirty-two of those carry no type yet.
Vlog stays a word.Why (his words): "We need to make sure that we also have creator content specific terminology correct … A vlog could also be a user story." The page had suggested letting it go; he overruled that. A piece that is both a vlog and a user story is covered by the August ruling that a piece may carry several types with one in the lead.
Event photo stays a word, and a photo should connect to the work it shows.Why (his words): "Event photo is good … it's more about how we managed the coverage of something to hide those relationships with the content, the event, the company, the products." The studio's own site gave the library both the installed animation and the photographs of it, and nothing connects them yet.
What these five answers change. None of the five words was on the list of 64 content types, so the list became 69 when the carve was regenerated the same day. Four placement calls follow that have not been asked: which group holds behind the scenes, vlog, event photo and infographic. System card's group is his answer above.
Ruled on how answers are processed and how questions are asked (11 September 2026)
Every sitting's answers are recorded the same session they arrive, and mapped, but never executed there.Why: a decision that is not written where the next page and the next session read it gets asked again. Recording means this page, the model page, and every page that teaches the term. Mapping means finding where the decided term lives in the code, the database, the prompts and the app, and adding that to the pipeline lane's list. The building happens later, in batches.
The building happens in Joe's order: how things relate and are stored first, then the prompts, then the pipeline, then the app's organisation and search.Why (his words): "part of my goal now is to ensure that we nail schema relationships, how it works, so we can nail prompts and nail sort of pipeline stuff, and then we should let a lot of that sort influence UI and organization of database and searchability." The walk runs in the same order, so storage questions come before naming and placement.
No question reaches Joe without a search of what is already ruled, and what the search found goes on the page.Why (his words): "We need to make sure every time we go to a new context window session that the model can somehow build the knowledge so it doesn't ask questions about something but it's like wait, the answer is right there, check it."
Ruled on how much of the review he answers (7 September 2026)
Every one of the 59 questions stays on the walk. None is skipped.Why: the day before, the idea had been to drop the questions that change no data — on a count of 25 that turned out to be wrong. Checked one question at a time it was 12, and the test itself was faulty: several of those touch no stored row only because the work behind them has not run yet, which makes them pending rather than unimportant. Of the seven that genuinely qualified, two were questions about how the product works that he would want to answer regardless. In the new layout the remaining five take about half a minute each, so skipping them saved a couple of minutes and left five loose ends. His call: drop the idea.
Ruled on how the review is laid out (7 September 2026)
The review moves to the hybrid layout: five decisions on one page, each answerable without opening anything, with the pieces and the numbers one click away.Why: the previous layout was correct and produced nothing — no decision came out of it in twelve days. Four layouts of the same decision were built so the choice could be made by looking rather than by reading a description. Joe walked all four: "okay i do love this version … i like this and would like to apply it across all open decision questions." Measured against the old pack, the hybrid puts the first answer inside the first screen rather than a page and a half down, and gives about 215 words per decision instead of 748.
Decisions are grouped into sittings by the KIND of question they ask, not by topic.Why: one page today mixes "should this word exist", "what should it be called" and "where does it belong", so the reader changes mental mode at every card. Five of the same kind of judgment is one layout learned once and repeated.
The words get rewritten in the same pass as the layout.Why: Joe has rejected this lane's writing twice — the console in August and the instruction box on this very page — and both times the format was fine and the language was not. His instruction both times was the same: read the Atlas documentation, which says things in plain English. A change of shape is the one moment when every sentence is being touched anyway.
The existing review pages stay until Joe retires them.Why: he has not walked the new surface yet. Deleting the old one before he has would leave nothing to fall back to, and the old pages cost nothing to keep while their generator is untouched.
Ruled while the review was being shortened (4 September 2026)
Seven stops fold into four, and two more come off the list — the review is 59 stops, not 68.Why: the pack was correct and too long, and he said so after reading the front door and the first decision: "its a LOT of words and certain parts make sense others are dense and kinda confusing." Five of the folds put a question and its echo into one question, because answering one already answered the other; three scheduling stops that only asked whether he looks at another page this round became one line on the front door. Nothing about a lean, a take, or what a surviving decision asks changed — a fold rewrites the question, it does not change the answer it is asking for.
A last-reviewed date on every model and decision page — do it, do not ask about it. Why (his words): "just do it". It was the only call on the agenda about noticing when a right answer has quietly gone stale, rather than about getting a word right once, so it is built rather than debated.
The site is renamed, and it stops opening on a landing page. Why (his words): "you can rename the site thats fine bc its also confusing bc it opens to this like landinga page which if you click "The model" in the top you think it would take you to that section of the docs but it takes you to that landing page id rather it just open on the actual docs page with sidebar we dont need an unnessecary landing page and make whatever exists now as the actual overview page in the docs so were not jumping in and out of the docs it feels like". This answers the naming question he had left open in August, in his own earlier words: "our atlas which we should really just rename as our brandtrackers docs… done whenever it makes sense". The title and the wording change now; the address it is hosted at is deliberately left alone until a promote, because changing it mid-review would break the only surface he reviews on.
Ruled at the consolidation round (17 August 2026)
The recommendation page is the one review surface. Why (his words): "its getting to be too many pages/places to look/review and understand what is supposed to be what." The shelves, the ruled-off list, the naming picks, the Retail & Commerce expansion, the Brand & Identity summary, and the digital-product answer all now sit on that single page, in reading order; depth pages are one click away and on the sidebar.
Every review page joins the sidebar, inside one collapsed group. Why (his words): "if you make a page but dont put it on the sidebar and it gets lost in the chat its like gone i cant find it." This amends the earlier two-entries rule on his own new words; the collapse keeps the sidebar from growing forever, and ratified pages still retire.
The pillar once called "Works & Groupings" is now "Groupings." Why (his words): "i still dont like works bc i never call anything really 'works'" — and the alternatives failed his own tests (three-kinds names grow a word per new kind; "Campaigns & Groupings" privileges one kind). Taken on his lean ("maybe Groupings"), flagged for his glance. The one-creative-many-appearances ruling is untouched; the page now says "one creative" where it said "one work."
Re-read everything, every round. Why: stale pages and dense walls kept surviving reviews. The writing standard now carries the loop rules: every round re-reads every page, every page is sidebar-reachable, and tables follow the short-cell rules on every surface.
Ruled at the second reconciliation round (16 August 2026, later the same day)
The identity shelf is named "Brand & Identity." Why (his words): "lets ensure we call it 'Brand & Identity'." The name matches the top level of his own brand-identity spreadsheet word for word. The shelf keeps its OPEN badge for depth — the visual / verbal / strategic carve-up is now in review, fed by the identity sources he named.
"Recruitment ad" is dropped from the ratified list. Why (his words): "recruitment ad just drop idk if we have any of those." The library was probed the same day: no piece carries any recruitment typing at all. The employer-brand meaning is not lost — it already lives with the employer-branding craft rulebook and the audience axes. This was also the one type that had no home under the five-worlds option. The ratified set is now 64.
"Offer" stays listed but leans remove — restated. Why (his words): "offer promotion feels like marketing intent or goal." The promotion-ness is a marketing intent, not a kind of piece; the pick confirms at the naming review. It is the only type left on the Still-Open shelf.
The identity vocabulary review is activated now, not after the carve closes. Why: he named the sources and asked for the extraction directly — his brand-identity spreadsheet, the GitHub Brand Guidelines 2025 book, the OpenAI guideline pages, the motion-identity framework, and a search for the rest. The review keeps two strata separate: kinds of piece (what the shelf holds) versus identity elements (what a piece shows or belongs to).
Ruled at the reconciliation walk (16 August 2026)
"Performance ad" leaves the content-type list. Whether a piece is paid or organic is a fact about its distribution — ownership, placement, ad format — never about what kind of piece it is. Why (his words): "performance ad, that's like organic vs paid, not a content type — wrong part of our ontology/schema/taxonomy." The industry's own mechanics agree: the same unchanged post runs organic and paid at once when boosted, so paid-ness cannot be a kind. The handful of pieces typed this way re-type to their real kind in the migration; paid-ness lands on the distribution facts.
"Public-service ad" is retired from the ratified list. Why (his words): "I don't know what a public service ad is — remove that." It was machine-proposed, never carried a live value, and no one could say what piece it names.
The "Advertising & Promotion" shelf is dissolved. Why (his words): "why advertising & promotion this to me is in campaign right?" An advertisement is not a separate family of work — it is campaign work, and the paid-or-organic part of it belongs to the distribution facts, which is the same reasoning that moved performance ad off the list. Its types re-shelve under Campaign and the other shelves his carve names; no piece changes.
Ruled at the reference review (11 August 2026)
Products get a model chapter of their own.Why: a product is an entity — that was already ruled — but the product layer carries enough structure of its own (detail records, a feature layer, direct links from content and campaigns, and the two-homes shape) that a passing mention on the entities page under-served it. The entities page keeps the ruling and points to the chapter; the chapter carries the full shape, including its honest gaps.
The top level of the craft tree is called a service; the rulebook level is called a craft.Why: the top-level lines — creative production, brand strategy, media strategy and buying — are what a client buys, and calling the whole tree "crafts" hid that. A craft is the specific, rulebook-carrying skill: the level where every entry says exactly when to tag it and when never to. The reference surfaces now say Services at the top; the middle level's name (group, discipline, or family) is the one part still open. Renaming anything stored in the database is separate build work.
Ruled at the ratification round (10 August 2026, later the same day) — recorded here on 12 September so it can be found
After the second review walk, Joe marked two proposed pages the same evening: the list of content types, and the channel stack (media type, then channel, then platform, then touchpoint, ruled at the second review walk below). The ruling below never reached this page. The search of earlier rulings found it on 12 September.
Every content type on the 10 August list was ratified that evening, and so was the channel stack.Why (his words): "if we are ONLY loking at the individual content types inside of proposed family i would say those are all good". He approved every individual type in that one sentence. Later rulings on this page removed some types and added others. He did not ratify the family names, the headings the types were grouped under: that evening he renamed two of them and set two aside, and on 13 August he re-opened the whole set. He approved the channel stack as it stood, then accepted every fix a completeness check proposed: missing channels added and four contradictions fixed.
Ruled at the second review walk (10 August 2026)
The follow-up packet came back answered. These bind; his exact words are in the banked response record. The open detail level under moments was answered after the initial pass: keep it open, then review and promote useful terms. That answer is recorded below.
When a specification and this decision surface disagree, this surface wins. The questions ledger and this page are the single record of what is settled; any other document that says otherwise gets corrected to match.
A campaign is a container, not a lens. The lens list loses "campaign". The grouping kinds are now stated: a campaign · a brand-and-identity program · a project · an event or season. A project is the deliberately flexible one — a retail experience with fifty pieces across in-store displays, or a genuine one-off — so that nothing is forced into "campaign" just because no other box exists. A campaign can itself contain events and, in time, news articles. Why: the fear was a rigid two-box world (brand-identity or campaign). The project kind absorbs the honest mess.
Vendor work is first-class, not a side effect. Case studies and portfolio write-ups scraped from the studios', agencies' and production companies' own sites are as valuable as the brand-side collection — "if we only scraped every studio, agency, production company's website and organized it, it would be just as powerful." The model and the pipeline plan must carry vendor-side collection explicitly.
All three scope declarations: IN. Offline media placement is modeled (a news article showing an out-of-home ad, an airport photo someone takes, scraped trade publications). Geography and language are captured. Reception is captured as signal — views, comments, their tone — without competing on ad-platform metrics; sentiment is noted as a future model of its own.
The channel is a stack, not a word. Media type at the top (digital, broadcast, print, streaming, outdoor, retail, physical and digital experiences), then channel, then platform (a specific service — one of the video or social platforms), then touchpoint (in-feed, a profile page, an inbox, a store display), with ad format and placement as their own fields. "Social" is a channel; a short-video platform is a platform; in-feed is a touchpoint. The earlier taxonomy workbook is the starting inventory; the derived six-value list in use today maps into the stack rather than surviving as-is. The full vocabulary is a proposal to review, not yet canon. (That was true at this walk; it was ratified later the same day — see the ratification round above.)
A content type is the industry-recognizable kind of piece, at the piece grain. Not where it ran, not why the campaign exists, not its role in the campaign, not the craft that made it. A launch film, a testimonial, a tutorial, a key visual, a logo animation — yes. A social post, a pre-roll ad, a "product campaign", a "hero" — no; those live on other axes (platform, placement, grouping, role). A piece may carry several types with one primary — the same pattern moments already use. The full master list is a proposal to review, not yet canon. (That was true at this walk; its types were ratified later the same day, but not the family names — see the ratification round above.)
The moments registry gets cleaned to one job. The browse level stays derived; the shelf list becomes browse shelves only, each with one group parent; whether something is offered as creative reference becomes a separate visibility label; the nested shelf is flattened. The vocabulary itself must serve two different readers — people who ask for narrative beats (logo animations, product interface moments, reveals) and production people who ask for shot types (establishing, close-up) — and typography needs sub-splitting (full-screen title animations, end cards with pricing or a call to action, opening titles).
Occasion is a first-class positioning axis, alongside audience and market.
A craft can have several parents. Character animation belongs under 2D and 3D animation at once — "maybe it's not as hierarchical, just relationship-based." One primary parent controls where it displays; the front-end flow is still to be worked out.
Content and Moments keep two pages. The combined page and its generator are retired. The distinction that settles it: is the whole piece this thing, or is there just a moment of it inside? A two-second title animation inside a film is a moment; a film that is title animation top-to-bottom carries it as its type.
The open detail level under moments stays open — harvest, then promote. The level keeps accepting new words (that is how a genuinely new technique gets named), but on a schedule the tail is reviewed: values used above a floor get promoted into a curated vocabulary with written definitions, and one-off spellings merge into their curated neighbor. The reviewer sees only the promotions, never the raw tail. Why: the open level did its job — the head of the list is genuinely useful — but 379 of its 577 values are used exactly once, and nobody should ever have to read that tail.
Entity identity, human-work survivability, and the taxonomy freeze are ruled as recommended: an entity is its kind plus its canonical name with aliases carrying variants, merges re-point rather than delete, and the pipeline stops minting entities without review · every classification records who said it, automation never overwrites a human row, and anything human work hangs on gets an identity that survives re-detection · the registry is the single governed surface, classifier instructions are generated from it, and every classification records the taxonomy version that produced it.
Ruled at the first review walk (4 August 2026)
Joe answered the open questions in one dictated pass. The rulings below are the ones that were clear enough to bind; his exact words are kept in the banked response record. Questions he asked to see more evidence for stay on the open list and get a follow-up packet.
A collection of work is a grouping, never a company. A campaign, a brand-identity program, a one-off project, an event or a season — each is a set of content items (and related facts) grouped together. It can have participants; that does not make it a company. Why (his reasoning): the whole idea of a content item is an atomic, singular piece that can be grouped many ways. The largest "company" in the system was really one of these groupings wearing the wrong label.
Owning a company and being someone's parent brand are two different facts, and the model must hold both. Microsoft owning GitHub is corporate ownership. Amazon over Amazon Prime, or iPhone over iPhone 16 Pro, is brand architecture — a house of brands with parent brands, sub-brands and product tiers. Depth is allowed in both; and a search for the parent must also find everything beneath it. Why: users want exactly that specificity — which tier is the premium one, what sits under what — and collapsing the two kinds of link into one destroys the distinction they are asking about.
A product does not "become" a brand — everything is an entity, and a product can carry its own brand identity while still belonging to its company. ChatGPT is a product with an identity of its own, distinct from how its company presents the whole family. The relationship map carries that nuance; no promotion rule is needed. Why: the question assumed a flip from one filing to another. The ruling is that both facts are true at once and the relationships express it.
Standing relationships between companies are part of the model. Who competes with whom, who is whose agency of record, who partners with whom — durable facts, recorded directly, across many kinds of relationship. Why (his words): "this entire product is a relationship map."
The empty buy-side is unbuilt pipeline, not a model flaw. The agencies and studios are missing because only one brand's content has ever been collected; filling them is part of how a brand gets onboarded, including a routing step for newly discovered partners (a new partner entity raises the question "should we onboard this one too?" — human-decided for now, automated later).
Research artifacts are content — captured, tagged, analyzable — but they are not creative work and never appear in creative browsing. Model cards, technical reports, white papers, press releases and third-party articles all belong to the record of a campaign or a company's output; none of them is what a person is looking for when they browse craft. Why: the product's promise is the brand-marketing creative. The ecosystem around it is context, kept close but never mixed into the shelf.
A moment is a video-only idea. Sections of a long page or regions of an image, if ever needed, will get their own word.
A company and a product position themselves differently, so they get separate positioning trees, both part of the entity's brand identity.
Time belongs on entities and their relationships. Ownership, type, name and standing relationships are dated facts, not single current values.
One piece of work, many appearances. The same film on the brand's site, on a video platform and in a press kit is one work; each appearance is kept with its own channel, engagement and surrounding information. Aspect-ratio variants are appearances of the same work too: the primary ratio leads, the vertical and square cuts are kept and findable, and the same file is never shown twice as if it were two works.
The model needs several pipelines, isolated from one another. Beyond the existing content pipeline: a brand-identity pipeline, an entity base-information pipeline (the URLs and links that define an entity's presence), and a campaign pipeline. A failure in one must not take down the others.
Live classification gets a daily recognition pass. On a backlog you can see a campaign cluster at once; day to day you cannot, so a recurring review looks at newly ingested content and asks whether it could belong to a campaign or another grouping that is only now becoming visible.
The timing layer serves the other pillars rather than standing alone. Publication time is captured on everything; shifts over years are the analysis the product exists to serve; signals and observations are derived from that record. Connect what makes sense — the architecture details are delegated.
The thin-but-canonical ownership chain is the real one. The link table the rules already name as canonical is right; its emptiness is unfinished work. Backfill it, and point every "this brand's work" surface at it.
The words we use
A thing in the world we track is an "entity". Not an actor, not a node — an entity, which is also what the data calls it. Why: one word, used by the product, the data and the people talking about it. Every extra synonym is a translation step that eventually gets translated wrong.
A "family" is what a thing IS. A "lens" is how the product slices it. The families are entities, works and groupings, content and moments, and concepts. Lenses are industry, audience, channel, occasion, craft. (Campaign was once listed here as a lens; ruled out 10 August 2026 — a campaign is a container you filter by, not a lens.) Why: these were being used interchangeably, which made it impossible to say where something belonged. The test is short: a family answers "what kind of thing is this?", a lens answers "how do I want to cut the shelf today?"
Audience, market position and channel are concepts, not containers. You classify a piece against them; you do not put it inside them. Why: a container is a thing that happened in the world and that you could point at — a campaign, an event, a project. "Instagram" is not a thing that happened. Treating it as a container invites a structure where the same piece has to live in several boxes at once.
Content
Where a piece appeared is derived from how it was collected, not judged by a model.Why: the source knows. A page pulled from a company's newsroom is newsroom content because of where it came from, and no amount of reading the text improves on that. This replaced a version where a model guessed, which was both less accurate and more expensive.
A content item is one piece of work a person could open and look at. A film is one. A landing page is one. A campaign is not. Why: it gives a single, checkable test at the boundary, which is what stopped the unit drifting between "a file", "a URL" and "a thing we published".
Moments
A moment is one beat inside a video, and belongs to exactly one video.Why: moments exist so a team can ask for the bit where the shoe turns in the light rather than for a whole film. Anchoring each one to a single parent keeps that promise simple.
Moments are described at three levels: eight groups for browsing, a fixed set of shelves for what the beat is doing, and an open level underneath for the specific craft.Why: the top two are fixed so that browsing is stable and so a classifier has a target that does not move. The bottom one is open so a genuinely new technique can be recorded rather than forced into an existing word. The later review settled how to maintain it: keep accepting new detail, then review recurring terms for promotion into a defined vocabulary.
A moment may sit on several shelves, and exactly one is its main shelf.Why: a closing card that also contains an animated logo genuinely belongs to both, and forcing a single choice loses real information. The main shelf exists so there is always one answer to "what do we call this and where does it sort?"
Some shelves are catalogue rather than browse. Catalogue describes browse visibility; the moments remain kept and searchable. Why: the August review separated visibility from classification and explicitly valued both production shots and narrative beats. An establishing shot can be useful to a production team, just as a logo animation can be useful to a design team. The exact default presentation is being clarified separately; a catalogue label does not establish that a shot has no reference value.
The moments taxonomy is frozen between planned versions.Why: you cannot train a classifier toward a target that changes weekly. Changes go through a version bump and a planned re-run, never a live mid-week rewrite.
What the model is designed against
We design for the shape we are scaling to, not around the gaps in today's data.Why: almost everything that looks thin right now is thin because the pipeline that would fill it has not been built. Only one pipeline exists, and it handles content. "Hardly anything is tagged to a campaign" is a statement about missing machinery, not evidence that campaigns are the wrong idea. Designing around that would encode a temporary gap into the permanent structure — and the structure is the expensive thing to change later, while the data is the cheap thing to backfill.
This is the whole reason this documentation exists rather than another round of fixing rows. The conversations it replaces sounded like "careful, there are ghost rows, you'll break the wiring" and "well, campaigns are barely tagged" — both true, both about work we already know we have to redo. Getting the architecture right for a complex thing at scale matters more than patching something we have already decided to rebuild.
How to apply it: when a check on the conformance report fails, first ask which of three things it is — the model is wrong, the build is wrong, or the work simply has not been done. Only the first two are defects. The third is a to-do list, and it must never be used as an argument against the model.
How this documentation itself works
Model pages are written in plain language for someone who has never seen this project, and a linter blocks internal codes, raw database identifiers and file references on them. Why: the model has two audiences, a person and a machine, and shorthand fails both. It also removes the excuse that a page is "technically correct" while being unreadable.
Relationship maps are diagrams written in the page source, never components.Why: the machine-readable version of this site is generated from the page source. Anything rendered by a component would be invisible to half the audience.
Worked examples are generated from live rows, not written by hand.Why: an invented example is unfalsifiable. A real one can be looked at and called wrong, which is the only way an error gets found.
Measurements never appear in model prose. They live on the generated state page; the model pages carry only the vocabulary and the rules. Why: a number typed into prose starts lying the moment the pipeline runs, and nothing catches it. Separating them means a model page can only be wrong about meaning — which is the one thing a human review can actually catch.
Where the data disagrees with the model, that is a check, not a paragraph.Why: prose goes stale silently, reads as an opinion, and nobody remembers to delete it once the problem is fixed. A check turns green by itself.
Ruled in July (10 July 2026) — recorded here on 11 September so it can be found
These rulings were ratified in July in the canonical classification model, but they never reached this page. As a result, the question they answer was asked again on five walks. They are recorded here in plain words.
Content category is not above content type. The two sit side by side on the same piece.Why: the feeling that category sits above type comes from something else, the grouping a piece belongs to. A campaign holds many pieces, and each piece has its own type. "Feature campaign" names the campaign a piece belongs to, not a label on the piece.
The old category field is replaced by two separate things: marketing intent and content role.Why: the old field mixed several questions in one word. Marketing intent says why a piece exists — product marketing, brand marketing, research. Content role says the job a piece does — to announce, explain, prove or educate — and a piece can do more than one. Whether a piece belongs to a campaign is a link to that campaign, never a word on the piece.
Joe's five output pillars are ways of browsing the library, never a label stored on a piece.Why: the pillars — brand identity and systems, campaigns, content, digital products and experiences, physical products and experiences — describe how people look across the work. A piece does not store which pillar it is in; the pillar is worked out from what the piece is and what it belongs to.
What contains what is a link between the two things, never a word written on a piece.Why: the July model keeps four kinds of fact apart: what contains what, the relationships between companies, the words that describe a piece, and the views people browse. A piece belongs to its campaign through a link, and a moment is part of its video through a link. Found on 11 September by the search of what is already ruled, before the foundations sitting asked how a piece inside another piece links back.
Ruled in July (20 July 2026) — recorded here on 11 September so it can be found
This ruling was made on the vendor case-study cards in July and never reached this page. The search of what is already ruled found it on 11 September, while the foundations sitting was being written.
Every piece gets one tag for what it is to us: the finished work, material showing how it was made, a photo or video of the work out in the world, information that belongs to a campaign, or a vendor's list of services.Why: a page of a model card was being judged as if it were a piece of design. Joe marked the card approved and asked for it in plain words: a photo of the work in the world means the event floor, a storefront, the app in someone's hand. The five values are registered, and nothing writes them to a piece yet.
Ruled in July (11 July 2026) — recorded here on 12 September so it can be found
On 11 July Joe signed off a set of one-page decision cards that followed the July classification model, ratified the day before. These four rulings came out of that sign-off and never reached this page. The search of earlier rulings found them on 12 September, before Joe answered the September questions about how pieces connect.
What a piece is, the job it does in its campaign, and which version it is are three separate facts.Why: a launch film is a type of piece. Being the hero or a teaser is the job it does in its campaign, so that job sits on the link between the piece and the campaign. A shorter cut or a vertical cut is a version of the same film, never a new type. That way a search for every hero launch film is simple: the type, plus the job in the campaign. He adopted it on condition that it is tried on a small set first.
A piece never stores "brand campaign", "product campaign" or "feature campaign". That is the type of the campaign it belongs to.Why: teams think in brand, product and feature campaigns. That grouping is the campaign itself: the campaign records its type, and its links to products and features narrow it. Why a single piece exists is its marketing intent, a separate label on the piece.
In the list of industries, "category" means the product market something competes in.Why: the list has three levels. A sector is a broad slice of the economy. An industry is a group of companies inside a sector. A category is the product market, and each company maps to the categories it competes in, with one marked as its home. Level names must be real market words. Products also carry a list of product markets, and it must become one list with this one or be mapped onto it.
Where a piece was seen is its own record, and one piece can have many.Why: where we collected a piece is not the same as where people saw it. Each place it appeared keeps its own channel, platform, placement, web address, dates and evidence. These records fill in as evidence arrives, not all at once.
Ruled in July (13 July 2026) — recorded here on 12 September so it can be found
On 13 July Joe approved the short dictionary of what each layer of the classification means, on condition that its rules can be seen and tested in the taxonomy admin page. It gathered rulings he had already made. These four were in it and never reached this page.
A four-way test decides where a new idea goes.Why: if it is a narrower kind of something on one list, it belongs in that list. If it describes something, it is a label on the thing. If it says one thing belongs to or is part of another, it is a link between the two. If it is a way people browse, it is a view, and nothing is stored for it.
A label on a piece must be true of the whole piece. A beat inside it is a moment.Why: the test is whether you would tell a buyer the whole piece is this. If yes, the label goes on the piece. If it is a beat you would clip and point to, it is a moment. A label on the whole piece must never be the only record of something true of one beat.
A search for a craft shows work that is that craft from start to finish.Why (his example): a search for motion graphics must show work an animation studio would show as its own. A live-action film with a short title animation inside it shows up in the moments results instead, not in the content results.
A moment takes its brand, product and campaign from its video, and adds only what is true of the beat.Why: the video already carries those facts, so the moment reads them through its link to the video. It adds its own labels, such as the kind of shot. It never changes the video's labels.
Ruled in July (17 July 2026) — recorded here on 12 September so it can be found
On 17 July Joe marked the decision cards on the market and craft lists. Smartwatches sat in two places, and his card said keep both; the follow-up round that day settled it as below.
A thing is stored once, and a view can show it in more than one place.Why: smartwatches belong under consumer electronics, and someone browsing watches should still find them. So the watches view includes smartwatches without a second copy of them.
Ruled in July (20 July 2026, the relationship cards) — recorded here on 12 September so it can be found
On 20 July Joe marked the cards on names and on how groupings are stored. These two rulings never reached this page.
One idea gets one name everywhere: on screen, in the stored data, and in the words the AI is given.Why (his words): "ensure we are consistent in naming across display tables schema prompts etc i dont want an inconsistency in how things are named at this point we can fix this now". A name that is wrong gets fixed now, even when that means renaming a stored table or column; it is not worked around.
Campaigns, brand-and-identity programs and projects are kept in one list, and each one says what kind it is.Why: they all hold pieces the same way, so one list holds them. The flexible kind is called a project; the decision record gives that as his pick over "brand program", and his own note on the card asked for whichever name is best practice. Events and seasons keep their own records and link to campaigns.
Ruled on the foundations sitting (3 October 2026)
Headings belong on the list of content types, not separately on each piece.Why: Joe said, "I'll take your recommendation on the heading." The sitting recommended keeping the heading on the list. A heading helps people browse types; a particular piece's purpose and membership remain separate facts. This settles where headings live, not every heading name or the main navigation.
Every captured document page stays accessible inside its document; selected useful parts can also be browsed independently and link back to it.Why: asked whether useful charts, designed graphics and other useful parts should be separate library pieces while all captured pages remain accessible inside the document, Joe answered, "Yes—select useful parts for independent browsing". This replaces the sitting's blanket suggestion that every part becomes an independent piece. A plain report page need not become creative reference. Research stays findable, with the existing separation from creative browsing.
A photograph of work links directly to the piece it depicts and keeps its own identity.Why: Joe said, "Yes, it links straight to the piece." The photograph and the depicted work answer different questions and both remain available. Where the work appeared stays a separate placement record, as already ruled in July. An uncertain target remains unresolved rather than being guessed.
Content types, groupings, type-list headings and distribution remain separate concepts; no mandatory content-category parent sits above them all.Why: Joe answered, "Yes—use these separate concepts", when shown the proposed separation. A logo, launch film and model card each keep their kind; a campaign, identity program or project can gather them. Channel, platform and touchpoint describe distribution. Pipeline routing uses available source and media facts rather than requiring campaign membership. Heading names and the exact routing implementation remain separate reviews. This settles the remaining foundations question without approving a migration.
Ruled in July (21 July 2026) — recorded here on 3 October so it can be found
The marked July cards and their original recommendations settle the product relationship that a later review card asked about again. These are recovered approvals, not new answers to the current walk. Approving the July rule did not move any records or establish that the database already enforced it.
A product has one identity and one family tree in the entity graph; its product-specific details attach to that identity.Why: Joe approved the recommendation that the entity graph carries brands, sub-brands, product lines and models, with a one-to-one product-detail record. A second product family tree and a separate link for the same product's brand identity were to be retired through implementation work. His qualification required precise product and feature types, market relationships, a family search that includes its descendants, and the ability to narrow to individual models. The later August ruling keeps company ownership distinct from parent-brand relationships; one graph does not mean those relationships are interchangeable.
A product family and a product bundle are different relationships.Why: Joe's approval called for family, model and version distinctions, with price and market tier described separately. He explicitly distinguished Microsoft 365 including Word, Excel and PowerPoint from a family of models or versions. The July recommendation also proposed a three-level cap and version attributes. At the August walk he asked to allow appropriate depth, so that earlier hard cap must not be restored as an unconditional current constraint. The detailed version representation still needs reconciliation before implementation; the family-versus-bundle distinction does not.
Ruled at the platform review (10 August 2026) — recorded here on 3 October so it can be found
The original platform example and Joe's explicit confirmation were already kept on the channel-stack page. This entry makes that approval findable in the shared decision record; it is not a new answer to the walk.
A platform links to its company when that company is tracked; where work ran and who made it remain different facts.Why: Joe confirmed the worked example as written: “one company, two roles, two tables, one link.” A company can advertise its own product on its own platform. The link lets someone find everything that ran on the service separately from everything its company made. The relationship is approved; recording it does not establish that its database implementation exists.
Ruled on vocabulary naming (16 August 2026) — recorded here on 3 October so it can be found
Vocabulary display names use Title Case, and slash-separated alternatives must be resolved into clear names.Why: Joe asked for “casing the first letter of each word” and rejected labels such as “commercial / spot” and “offer / promotion.” The naming review must determine whether each pair means one thing with an alias or two different things. This does not ban meaningful multiword names such as Corporate Communication, and it does not settle the still-open naming choices or authorize a blanket replacement of “film.”
Ruled on AI-generated craft (17 July 2026) — recorded here on 3 October so it can be found
Joe's marked July card approved Generative AI Production and the qualified rule below. The original card explicitly preserved evidence of human craft as an exception. This recovers an earlier approval; it is not a new answer to the current walk or an instruction to remove existing classifications now.
AI-generated work has a Generative AI Production craft home; photographic appearance alone does not justify cinematography or live-action craft claims.Why: the approved card kept origin separate from craft and required removal of unsupported camera-craft claims from generated work unless human craft was evidenced. A generated image can look photographed without a shoot. The rule does not erase other evidenced craft or turn an advertisement about AI into AI-generated work. Existing whole-piece and moment rules still apply. Later service-tree arrangements must be respected; recovering this meaning does not restore the original card's old parent placement or execute a migration.
Ruled on strategy evidence and compact display (20 July 2026) — recorded here on 3 October so it can be found
The marked alignment cards explicitly approved these two recommendations. Their historical database counts and implementation claims are not carried forward as current facts. These recover earlier approvals; they are not new answers to the walk.
A source's documented strategy claim stays distinct from our own inferred analysis of the work.
Why: Joe approved the case-study routing recommendation and emphasized collecting studio and agency sources alongside the work they explain. The approved route keeps the documented claim linked to its source, work and credited maker; our model's interpretation belongs with derived analysis. Who recorded it and how it is supported remain separate facts. An explicit source statement is not automatically verified truth, and scraping a page does not turn an inferred conclusion into a source statement. The exact original routing is retained in the evidence record for implementation checking; this recovery does not certify the collectors, database or readers.
Compact cards limit the labels they display, while search and filters retain all applicable classifications.
Why: Joe approved a card showing its content type and one primary craft with a count of other crafts; a detail header showing the primary craft and two others; and the full classification table in the detail panel. This prevents crowded cards without throwing away valid classifications. The same recommendation withheld raw confidence scores before calibration. These display budgets do not decide how many whole-piece content types to store, promote moment-only classifications onto the whole piece, or certify an existing layout. They are the approved starting point for the shared layout review.
Ruled on craft meaning, naming and review (17 July 2026) — recorded here on 3 October so it can be found
These approvals come from the marked July cards and their original recommendations. Later clarifications are kept alongside them. Their old counts, tree placement and execution plans are not evidence of today's implementation, and none of these entries is a new answer to the current walk.
A supported broad craft classification is valid when its specialization is unknown; a fabricated General label is not.
Why: Joe approved keeping a classification at the known discipline level when the narrower specialization could not be established, and retiring General on that basis. This was not permission to classify at any top-level service. Later discussion left the middle tier's name open; the honest level of certainty is the rule being recovered here.
An unresolved required main classification or a classification below its confidence threshold goes to human review.
Why: Joe's approval explicitly required uncertain main labels to be flagged and rejected a general junk drawer. Evidence for a broad craft can still justify that broad classification; that is different from inventing an answer to a required question. This approval sets no numerical threshold and does not make every descriptive list require a single primary value. The separate display ruling withholds raw confidence scores before calibration.
A craft names a deliverable; a treatment that restyles other crafts' work remains a separate description even when it is hireable.
Why: Joe ratified the output-versus-treatment test. Later approvals kept physical stop-motion work distinct from an AI-generated stop-motion look and allowed the same name on different lists when the meanings differ. The old combined execution bucket must not be restored by copying its examples. A commercial's format alone does not establish commercial-production specialization.
Audio Production names the broader field; Sound Design remains a distinct craft within it.
Why: Joe approved renaming the broader Sound Design heading to Audio Production while retaining the narrower Sound Design craft. This avoids making the parent and child mean the same thing. It does not restore the old service placement or settle the later open name for the middle level.
Logo Design, Brand Identity Design and the wider identity program describe different scopes of work.
Why: Joe approved Logo Design for the mark and separately approved Brand Identity Design for the design work across the visual system. The whole identity program can include verbal, strategic and other creative work as well. The design craft must not absorb the entire program, and the broader program does not remove the legitimate design craft.
Being a video is not sufficient evidence for a cinematography claim.
Why: Joe approved reviewing the classifications rather than merely merging their names, warning against everything appearing under Cinematography. On 21 July he explicitly approved moving cinematography and direction to credits. Preserve that later placement instruction rather than reviving the earlier browse branch. The separate AI-craft ruling also requires evidence of human camera craft; no migration or current credit implementation is claimed here.
Industry names convey useful detail about the narrower market instead of merely repeating the sector name.
Why: Joe approved descriptive names that help someone move from a broad industry to the kind of work they want. This recovers the naming principle, not every individual July label, a particular number of levels, or the separately unverified rule against all shortened words.
Photography names include Photography, and compact labels carry the context needed to understand them.
Why: Joe approved both self-describing photography names and contextual chips. A name should still make sense when shown away from its parent. This does not freeze the original photography inventory, the old dimensionality values, or a specific visual arrangement of chips.
Ruled on prompt quality and reanalysis (17 July 2026) — recorded here on 3 October so it can be found
Classification prompts use shared vocabulary definitions and detailed examples that explain what to choose.
Why: Joe approved the registry consistency guard, asked for definitions managed in one place for people and machines, and explicitly preferred comprehensive prompts with examples over vague instructions. The vocabulary supplies the allowed choices; the instructions explain their boundaries. This does not mean current prompts are already generated correctly, or that longer text alone produces better classifications. Verify the composed prompt, output validation and results together.
Reanalysis of affected work is an accepted part of changing vocabulary; consistent classifications take priority over preserving old runs.
Why: Joe accepted rerunning work as meanings are retired or replaced, emphasizing consistency while the product is still being developed. This recovers the direction, not a standing instruction to rerun the entire library after every edit. Determine which work is affected, preserve human corrections, and verify safe replacement before a run. Current cost and activation gates still apply. Historical acceptance is not evidence that a rerun has happened or that its output is correct.
Ruled on dimensional browsing and market tier (19 July 2026) — recorded here on 3 October so it can be found
These are recovered marks on the July recommendations. Later craft-parent qualifications remain in force. The old cards' storage descriptions and rollout plans are not evidence of current implementation.
A piece described by both 2D and 3D is findable through either description without duplicating the piece.
Why: Joe approved the worked motion-graphics example: one piece, descriptive chips and searches that find it through either dimension. He later agreed that a craft can have several parents. Preserve both meanings: a piece's dimensional description is distinct from the relationships between craft names. Do not restore the earlier restriction against dimensional craft parents, fix the exact storage design, or promote a moment's description onto its entire film through this recovery.
Product-market categories use plain product names, with market tier recorded separately and still available for browsing.
Why: Joe chose FOLD when the review contrasted preserving tier-worded category names with plain product nouns and a separate tier description. The Watches example explicitly retained luxury-versus-mass-market browsing. This recovers the naming and browsing requirement, not every old label, the exact tier vocabulary or where its relationship is stored. It is not permission to remove an existing browsing path before its replacement works.
Ruled on audience relationships (21 July 2026) — recorded here on 3 October so it can be found
A piece can address multiple audiences; those audiences must remain available for discovery.
Why: Joe approved the multi-audience recommendation and emphasized teams studying enterprise and developer audiences. The old card also proposed deriving or removing a supposed duplicate goal field. Inspected prompt definitions, stored examples and a preservation amendment distinguish who a piece addresses from what it wants them to think, feel or do. That bundled retirement premise must not be carried forward. This recovery approves no deletion or rename and does not certify classification quality or complete filtering.
Audience relationships are required across companies, products, features and content.
Why: Joe explicitly extended the audience requirement across these subjects. His earlier onboarding notes asked for a company-and-product map with target audiences and suggested an agent's first pass followed by human review. They left the pipeline placement uncertain. Preserve the requirement without treating that suggestion as a settled orchestration design, fixed audience hierarchy, automatic inheritance rule or approved schedule.
Ruled on browsing, versions and useful labels (4 October 2026)
Joe answered the browsing sitting and clarified the remaining distinctions during the planning review. These are approved meanings and product requirements. Recording them does not certify their database, pipeline or interface implementation.
One library offers visible, guided entrances into companies, products, categories, audiences, disciplines, services, channels and groupings.
Why: Joe wants people to understand what they can research without already knowing the vocabulary. Companies should introduce brands, agencies, studios and production companies before requiring someone to search a complete directory. Useful subcategories lead toward results; precise search can bypass that orientation. The same evidence can be reached through several relationships without copying the work or forcing one hierarchy. Compact expandable navigation, pinning and restrained pill controls are requirements to demonstrate, not a final layout choice.
Distinct cuts, languages and layouts are separately browsable versions of a work; publications and captures remain separate.
Why: Joe confirmed linked versions for the launch-film example. The shorter cut can be compared with the longer cut. The identical shorter cut published in two places does not become two creative works. Captures preserve what was collected from each publication. Connections need evidence; a matching title is insufficient. This clarifies the older use of “appearances” for both versions and distribution without approving a physical storage design.
Several types can describe a whole piece, with one leading; a contained beat never supplies a whole-piece type by itself.
Why: Joe confirmed that a film which works throughout as both a customer story and a product demonstration can be found through both types. A brief logo animation belongs to its moment. The parent film remains reachable through that matching moment, with the reason for the match visible. It must not enter whole-piece logo-animation results merely because it contains that beat.
Moment filters use maintained choices with definitions; rich descriptions can propose new terms for review.
Why: Joe confirmed reviewing new terms. Descriptions preserve what makes the reference useful without silently creating canonical filters. The exact vocabulary remains a separate review. This reinforces the earlier open-detail and promotion decision; it does not freeze the current prompt choices or approve a promotion list.
Ruled on the ongoing documentation review (4 October 2026)
Before a new review sitting is presented, reconcile the documentation with recorded decisions and verify the hosted site.
Why: Joe asked for the actual documentation site to remain canonical. The cycle includes an independent assessment of taxonomy, relationships and architecture, checked against the real data path. Current pages lose superseded guidance; dated source material and decisions keep the history. Coverage and unverified claims stay explicit. A model decision, an implemented constraint and a successful user journey are different outcomes and need different evidence.
The site is named BrandTrackers Docs. The existing address remains in use. First-pass vocabulary definitions remain proposals until reviewed; rewriting them in plain language is not approval.
Ruled on useful differences (4 October 2026)
Joe answered the five distinctions and clarified the campaign wording in chat. These decisions settle meanings. Schema, prompt, pipeline and interface implementation still require separate verification.
User Story is the shared name for accounts of using a product; Customer Story is an alternative search name.
Why: Joe accepted one name and chose User Story because use does not require a customer relationship. The label does not assert payment or purchasing status. Keep the subject and evidenced relationship to the product separately. Boundaries with Case Study, Testimonial and Product Review still need definition examples.
Endcaps remain distinguishable from other aisle displays, with a shared way into in-store displays.
Why: Joe accepted the recommendation and reinforced the photograph-of-work relationship. His OpenAI Dev Day example describes installation photographs and an original full-frame animation collected from Studio Dumbar. Both must stay accessible together without becoming the same piece. Preserve the direct depicted-work relationship when evidenced and the shared event or project context. The source example is his reported use case, not a newly verified set of database links. Exact type-versus-placement choices remain for the retail review.
A branded object keeps its kind, with gifting recorded separately when evidence supports it.
Why: Joe accepted the recommendation. A hoodie can be sold or given away; its photograph does not establish which. A useful gift set may be browsed as a whole and linked to its parts. This does not authorize an automatic rename or deletion before the gifting fact has a verified home.
How a product appears and the role it plays in a moment remain separate questions.
Why: Joe explicitly accepted the revised recommendation. An interface can be shown while the product is the focus, actively used or incidental. Overlapping words do not establish identical meanings. The exact vocabulary and unexpected stored values still require review and writer tracing; no conversion or reanalysis is approved by this answer.
A campaign may continue without a planned ending; duration and purpose remain separate descriptions.
Why: Joe clarified: “It can go on without a planned ending correct”. He also required ongoing monitoring, review and tracking. Evidence must still establish a coordinated body of work. A saved search does not become a campaign, and missing dates alone do not prove an ongoing campaign. This is a workflow requirement, not approval of a schedule or activation of automation.
Ruled on campaign membership (4 October 2026)
Campaign membership replaces Campaign Film as a content type.
Why: Joe explicitly agreed with this recommendation in the film and advertising sitting. Being part of a campaign describes a relationship. Preserve that relationship and the evidenced role of the piece within the campaign. This amends the earlier approval of the name; it does not approve removal of campaign information or require every video to have a known campaign. Registry, routing, saved searches and existing assignments still need coordinated implementation checks.
On 5 October, Joe answered the four follow-up choices. The later decisions replace the earlier pending status while preserving the original sitting as history.
Ruled on earnings releases and annual reports (4 October 2026)
Earnings Release identifies a company’s announcement presenting financial results and remains findable with press releases.
Why: Joe answered “Use Earnings Release as recommended.” The approved boundary is the announcement that actually presents the results. A notice linking to it, an earnings-call recording and an analyst's article keep their own forms and connections. Corporate Communication is its accepted browsing group. This replaces the earlier Earnings Report candidate wording without relabeling existing material automatically.
Annual Report identifies an organization’s account of a year; Investor Report is not automatically its synonym.
Why: Joe answered “Use Annual Report as recommended.” The report can be a website, PDF or captured print document; elaborate design and public-company status are not requirements. Investor material can instead cover a quarter, an event or another subject. Corporate Communication is its accepted browsing group. Original titles remain preserved.
These are approved meanings and placements, not database changes. Captured pages and useful independent parts retain their previously agreed relationships. Research remains findable without filling default creative browsing. The review record preserves the checked examples. The wider heading review and implementation audit remain open.
Ruled on reports, article structure and reaction videos (4 October 2026)
Trend Report is a distinct type; it is not automatically another name for Research Report.
Why: Joe answered “Keep Trend Report as recommended.” The report identifies and explains patterns or changes in a field. Research Report remains approved, and a piece can qualify as both when both fit. Neither label guarantees reliable conclusions. Preserve original titles and sources; do not automatically fold these reports into White Paper.
Listicle describes a searchable list-based article structure alongside the piece’s type.
Why: Joe answered “Use searchable structure as recommended.” The list must organize the article; a few bullets are insufficient. Advice, comparisons and editorial articles can share that structure while retaining their own types. This settles the distinction, not which database field stores it or whether the current Writing Style vocabulary is suitable.
Reaction Video is a type when a presented response organizes the whole video.
Why: Joe answered “Use Reaction Video as recommended.” A presented response to source material, an event or a product must be central to the whole video. A brief reaction shot remains a moment, and a product review is not automatically a reaction video. Channel and sponsorship stay separate. Link identifiable source material only when evidence supports the connection.
These three answers amend part of the earlier bundled question. Corporate Film remains open in the video discussion. Browsing headings, registry reconciliation, prompt rules, correction handling and search implementation remain separate work. The review record keeps the examples and original alternatives.
Ruled on interviews and learning material (4 October 2026)
Interview is a type; Q&A is a searchable presentation, not an automatic synonym.
Why: Joe answered “Interview type + searchable Q&A presentation (Recommended)”. An Interview is organized around eliciting answers, experience or views. Q&A presents questions and answers and can describe an interview, an audience question session or a FAQ. A FAQ is not automatically an interview, and a brief interview does not classify its whole parent.
Course and Lesson keep distinct meanings, with evidenced membership and order.
Why: Joe answered “Keep Course and Lesson distinct (Recommended)”. A Course is an organized learning resource or program. A Lesson is a focused instructional unit that may stand alone or belong to a course. Preserve evidenced order, reuse and original titles without inventing prerequisites. A course landing page does not prove that every lesson was captured.
These answers settle two parts of the earlier naming batch. Other pairs remain open. Exact storage, browsing headings and implementation require separate verification. The review record preserves the examples and alternatives.
Ruled on source and subject browsing (4 October 2026)
Collection-source products and products shown or discussed have separate, evidenced browsing paths.
Why: Joe answered “Separate browsing paths (Recommended)”. People can ask for material collected inside an app separately from material showing or discussing that product. An app capture could discuss other products; neither relationship proves the other. Keep evidence for each, and leave unknown relationships unresolved. Collection source does not establish who made the work, which tool made it or which campaign it belongs to.
The review record preserves the example and alternative. This approves the meaning and browsing direction. Storage, classification writers, search, tracker rules and correction handling remain for implementation verification. Whole-page membership and useful parts retain their existing approvals. The later retail decision also settles Product Detail Page as a whole-piece type; its earlier pending status is superseded.
Ruled on contribution scope (21 July 2026) — recovered on 5 October
Contributor relationships can identify particular moments as well as whole pieces and groupings.
Why: Joe approved making credits a core part of the model and explicitly described different agencies, studios and people contributing to a campaign, its pieces and the animation moments inside a video. His Buck example expressed the required relationship; it is not a newly verified attribution. A whole-piece credit must not be treated as evidence that a contributor made every particular interval.
This recovers an earlier answer. The new question about allowing or deferring moment credits repeated that approval and is withdrawn. Exact role names, whether any roles need a separate moment-level vocabulary, interval anchoring and pipeline support remain to verify. No new answer or working implementation is claimed. The corrected review keeps the scope and evidence limits visible.
Ruled on Format (21 July 2026) — recovered on 5 October
Format records whether a piece is image, text, video, audio or interactive; how it was made and where it appeared are separate facts.
Why: Joe approved one canonical Format and qualified his answer with “assuming confirming this is like image video etc”. The original recommendation separated production descriptions and collection details from that question. The later channel-stack approval uses Media type for distribution. It does not replace Format with a production technique or a distribution label.
The old proposal to rename the entire mixed list as Production Form is withdrawn as a standalone naming question. Its values do not all answer how something was made: Static Image and Audio Only describe media kind, while Live Action and Motion Graphics describe production or presentation. Retaining the useful distinctions requires checking each meaning, not just changing the heading.
This recovers the approved distinction. It does not certify a lossless mapping, a completed retirement, or support for every format. Routing, prompts, stored assignments, human corrections, saved searches and all readers still need a coordinated implementation contract. The evidence note preserves the original recommendation and the limits of the current checks.
Ruled on shots and moment descriptions (5 October 2026)
Joe accepted these explicit recommendations in his morning reply. Exact definitions and implementation remain separate work.
Shot Types remains the group for Dialogue and Establishing.
Why: Joe accepted keeping the shot group. A logo animation keeps its own shelf. This preserves a home for production-shot research without forcing every moment into a shot type.
Shots has a visible browsing entrance that includes matching production shots.
Why: Joe explicitly accepted a visible entrance. People should not need a hidden catalogue switch to find matching shots. Featured selection remains separate from eligibility in the chosen research path.
Camera framing stays separate from scene activity and story descriptions.
Why: Joe accepted the separation. A close-up can show a reaction during an interview; those descriptions must not erase one another. His additional shot examples feed the exact vocabulary review, not an automatic rename or merge list.
Ruled on chat and analysis (5 October 2026)
Joe accepted these explicit recommendations in his morning reply. Exact definitions and implementation remain separate work.
Chat Interaction stays broad, with evidenced AI participation separately searchable.
Why: Joe accepted the recommendation to distinguish a chat exchange from its participants. Unknown participants stay unknown, and a title mentioning AI does not prove that an AI generated a response.
Data Analysis describes the task, with AI involvement recorded separately when supported.
Why: Joe accepted separating the analysis task from AI involvement. Translation by an AI is not automatically Data Analysis. A brief activity describes its moment rather than the whole video.
Ruled on videos, advertising and offers (5 October 2026)
Joe accepted these explicit recommendations in his morning reply. Exact definitions and implementation remain separate work.
Brand Film and Product Film leave the canonical type list; preserve form, campaign, subject, role and source wording.
Why: Joe explicitly approved removal and the campaign example: related launch video, key visual and retail work can be explored together. Unknown campaign membership stays unresolved. Launch Video wording, Corporate Film and other names are not silently resolved by this approval.
Commercial and Promotional Ad leave the canonical type list; advertising evidence and source wording remain searchable.
Why: Joe explicitly approved removal. Preserve source descriptions, supported advertising intent and individual advertising appearances. Promotional language alone does not prove a paid placement or sponsorship. Do not replace the removed names with Advertisement Image or a vague Marketing Video type.
Sponsored Content is searchable through evidenced relationships and disclosures alongside the piece’s form.
Why: Joe accepted the relationship approach. Preserve the disclosure, participants and covered work or publication. A general event partnership does not make every associated piece sponsored, and paid amplification alone does not prove creator sponsorship.
Offer leaves the whole-piece type list; requested action, price, benefit, terms, subject and source location remain searchable.
Why: Joe explicitly accepted removing Offer as a whole-piece type after discussing its importance. Offer remains a research concept. A requested action, a price and what someone receives are separate observations that can coexist. They apply to images and pages as well as video, and need not occur in an end card. Keep wording, unknown conditions, subject and source location. A shopping demonstration can also show another product’s price; that is not the app’s subscription price.
Your additional requirements for feature research and later-arriving vendor work are carried into the Products and Groupings explanations. The review desk lists the remaining choices. These approvals authorize no database, prompt or automation change.
Ruled on retail, pages and shopping (5 October 2026)
Complete displays keep recognizable types, with location recorded separately.
Why: A complete endcap display and its constituent artwork remain distinguishable. A broad display description is valid when the evidence cannot support a narrower one. Exact retail types remain to audit.
Product Detail Page is a type for a complete page presenting a particular product for evaluation or purchase.
Why: Extracted product photography keeps its own form and evidenced product link. The page is the piece when researched as a whole and can be the touchpoint where another asset appears. A screenshot does not establish every page state or interaction.
Shoppable describes an evidenced interaction in a publication or experience, not a new whole-piece type.
Why: Keep the underlying piece and the available shopping action distinct. This applies across social, e-commerce and physical experiences. A photograph of a checkout kiosk or a downloaded video does not inherit the original shopping controls.
Ruled on identity names and motion (5 October 2026)
Brand Guidelines is the shared type for qualifying guidance; source titles and scope stay preserved.
Why: Brand Book, Style Guide and Digital Guidelines lead to this type only when their contents qualify. A strategy-only book or programming style guide is not automatically identity guidance. This is not a blanket alias conversion.
Key Art finds qualifying Key Visual work while preserving the creator’s wording.
Why: A principal visual for a brand, campaign or product can share the Key Visual type. Not every illustration or animation keyframe qualifies. Appearances and photographs stay linked but distinct.
Motion has a visible identity heading alongside the other identity headings.
Why: Motion remains part of brand identity and connects with visual, audio and digital expressions. A visible entrance does not require an exclusive hierarchy or a complete motion system for every brand. The exact elements and physical relationships still need verification.
Ruled on services and kiosk entrances (5 October 2026)
Use Service → Discipline → Craft for the three named levels.
Why: Creative Production is a service, Animation a discipline and Character Animation a craft. This names the explanation and browsing structure; it does not approve every current row, require only one parent, add an Execution tier, or equate company type with skill or credit.
Interactive Kiosk has a primary home under Interactive & Digital Experience and a second entrance under Retail & Commerce.
Why: Both entrances use one type and definition. A non-retail kiosk remains distinguishable; type-list placement does not establish retail use, duplicate pieces or classify a passive screen as interactive.
These eight choices were accepted separately in Joe’s reply and kiosk clarification. They approve meanings and browsing direction. The complete services vocabulary, credit roles, distribution mapping, schema and pipeline still need their own audit. The review desk shows what remains.
Ruled on stated strategy (5 October 2026)
Joe accepted all three displayed recommendations. These are approved meanings, not a claim that their storage, classification or research journeys already work.
Purpose, Mission and Vision remain searchable under the source’s names without forced interchangeable or mandatory slots.
Why: Preserve actual wording, source heading, subject and date. Any comparative interpretation stays separate; missing statements remain unknown.
Values describe stated principles and remain distinct from purpose statements and observed behavior.
Why: A source stating a value does not prove conduct or transfer that value to every product. Preserve each principle’s explanation and supporting passage.
Brand Promise and Value Proposition have distinct research meanings, with supported overlap allowed.
Why: A promise states a commitment; a value proposition offers a benefit or reason to choose. Preserve original terminology and supported subject and context. A feature alone does not establish its benefit.
The answered review preserves the examples and alternatives. The broader identity and strategy bundle remains partly answered. Preserve source evidence, subject, language and revisions; verify prompts, publication, search, trackers and Ask before claiming implementation.
Ruled on verbal identity (5 October 2026)
Joe accepted all three displayed recommendations. These are approved meanings, not a claim that their storage, classification or research journeys already work.
Brand Voice describes stated communication character and principles; Tone Variations describe contextual adaptations.
Why: Observed tone does not by itself establish a brand’s stated voice. Vocal performance is separate. Evidence can cover company, product or campaign context without automatic inheritance.
Key Messages describe intended ideas; Tagline System describes recurring short expressions and their usage rules.
Why: Keep source-stated messages, exact wording and inferred summaries distinct. A one-off headline is not automatically a tagline system; similar wording does not establish campaign membership.
Naming System describes rules for forming and organizing names; Terminology describes preferred words, meanings, spelling and usage.
Why: A list of names does not prove naming rules, and shared prefixes do not prove product-family or bundle relationships. Preserve aliases and evidenced scope separately.
The answered review preserves the examples and alternatives. The broader identity and strategy bundle remains partly answered. Preserve source evidence, subject, language and revisions; verify prompts, publication, search, trackers and Ask before claiming implementation.
Ruled on digital design systems (5 October 2026)
Joe accepted every displayed recommendation. These meanings do not certify implementation.
Digital Design System is the shared concept across supported platforms, while genuinely distinct systems remain distinguishable.
Why: Keep supported Web and App scope, versions and product relationships. A captured screen does not establish system membership.
Responsive Layout Rules remains a named, searchable description alongside Layout and Grid System.
Why: Preserve documented adaptation guidance separately from observed screen behavior. Keep the detail proportionate; no mandatory exhaustive breakpoint analysis is approved.
The answered review preserves examples and alternatives. Other entries in the original bundles remain open.
Ruled on visual identity elements (5 October 2026)
Joe accepted every displayed recommendation. These meanings do not certify implementation.
Themed Color Palette describes colors assigned to a supported context within an identity.
Why: Preserve the product, topic or campaign context without inferring membership from hue alone. This replaces the narrower proposed Topic Color Theme name.
Gradient System describes reusable gradient choices and their application rules.
Why: An observed gradient remains searchable without proving a reusable identity system. Preserve source edition and evidence.
Shape System describes reusable shapes and their construction or combination rules.
Why: Distinguish reusable shapes, guidance and finished applications. Similar geometry does not establish shared ownership or identity.
Brand Character System describes recurring characters and their identity rules; a mascot is one possible role.
Why: Preserve the named character, system and depictions separately. A one-off character, logo or licensed partnership needs its own evidence; no automatic entity or product creation.
The answered review preserves examples and alternatives. Other entries in the original bundles remain open.
Ruled on physical work and signage (5 October 2026)
Joe accepted every displayed recommendation. These meanings do not certify implementation.
Installation and Exhibition Booth remain distinct types, with supported overlap allowed.
Why: Installation describes deliberately arranged spatial work; Exhibition Booth describes an exhibitor’s designed visitor space. Booth Piece is not an automatic alias, and a photograph keeps its own identity.
Signage describes a designed sign or coordinated sign system, with retail use and location recorded separately.
Why: A photograph of a sign is not the sign. Keep independently useful signs linked to their system, with evidence for intended and observed use.
In-Store Digital Signage is a searchable presentation context, not an automatic type for every media file shown there.
Why: Preserve the screen presentation, component artwork, captures and evidenced place and time. A screen does not prove motion, interaction, payment or a complete playback sequence.
The answered review preserves examples and alternatives. Other entries in the original bundles remain open.
Ruled on design in physical spaces (5 October 2026)
Joe accepted every displayed recommendation. The repeated physical-space confirmation adds no duplicate decision. These meanings do not certify implementation.
Environmental Graphics remains a defined, searchable description of graphics designed for physical spaces.
Why: It can cover wayfinding, wall graphics and coordinated signage across settings. Location alone does not establish a commission, identity system or authorship.
Retail Design remains searchable for a designed retail environment or a substantial coordinated part of it.
Why: A retail location does not establish retail-environment design. Preserve each contributor’s scope and distinguish the coordinated environment from individual components.
Exhibition Design remains searchable for the arrangement of an exhibition or exhibitor space.
Why: Keep the space, structures, graphics, shown media and documentation connected but distinct. A component’s maker is not automatically the designer of the whole space.
Workplace Branding names the narrower identity description; wider Workplace Design remains separately available when supported.
Why: Branding can overlap with environmental graphics. Interior architecture, furniture and lighting do not become identity-system work merely because they appear in an office.
The answered review preserves examples and alternatives. Source, depicted work, contribution and system relationships need their own evidence. No schema, prompt or automation change follows from this approval.
Ruled on app icons and interface components (5 October 2026)
Joe accepted every displayed recommendation. The repeated physical-space confirmation adds no duplicate decision. These meanings do not certify implementation.
App Icon replaces the mixed App Icon Suite description; interface icon sets remain distinguishable.
Why: Preserve the identified application and supported platform or appearance variants. Favicon and Avatar are not automatic synonyms; a captured screen is not automatically Icon System work.
Interface Components remains searchable by component kind, with documented guidance distinct from observed examples.
Why: Button, Form and Navigation illustrate kinds, not a complete approved vocabulary. Do not create a whole-piece type for every fragment or infer behavior and system membership from a still image.
The answered review preserves examples and alternatives. Source, depicted work, contribution and system relationships need their own evidence. No schema, prompt or automation change follows from this approval.
Ruled on interactive discovery (5 October 2026)
Interactive & Digital Experience remains a shared browsing heading for interactive work.
Why: The heading organizes discovery without becoming a mandatory parent or a pipeline-routing rule. Captures and recordings remain distinct from the operable experience.
Joe accepted the recommendation on the review page. This settles the intended model; it does not certify capture support, storage, classification, search or creation paths. Those checks remain in the comprehensive audit.
Ruled on music labels (5 October 2026)
Music Label is a company type, not a separate basic kind of entity.
Why: Preserve the label identity and evidenced parent, artist, recording and business relationships. A label need not be separately incorporated; a credit alone does not prove rights or ownership.
Joe accepted the recommendation on the review page. This settles the intended model; it does not certify capture support, storage, classification, search or creation paths. Those checks remain in the comprehensive audit.
Ruled on podcasts and brand stories (5 October 2026)
Podcast Episode names a complete installment across audio and video; excerpts remain distinguishable.
Why: Preserve evidenced show and episode membership, source and media. Do not rename every stored Podcast Video as a complete episode.
Brand Story remains a type; Origin Story is a searchable narrative focus about beginnings.
Why: Origin focus can concern a brand, product, project or person. A passing reference does not classify the whole piece.
Joe accepted the recommendations on the review page. The approval settles meaning; record mappings, precise definitions and end-to-end implementation remain to verify.
Ruled on books, overviews and speaker cards (5 October 2026)
Book is a whole-piece type; eBook describes a searchable digital edition.
Why: Keep covers and download pages distinct. Equivalent file encodings alone do not establish different creative works.
One-Pager is a whole-piece type for a concise, standalone overview.
Why: Keep actual page count separately; a two-sided overview can qualify, while an extracted report page does not automatically qualify.
Speaker Card is a whole-piece type for a designed speaker introduction.
Why: Keep the finished card, reusable template, preview and photograph distinct, with evidenced speaker, event and publication relationships.
Joe accepted the recommendations on the review page. The approval settles meaning; record mappings, precise definitions and end-to-end implementation remain to verify.
Ruled on event recaps (5 October 2026)
Event Recap is the type for an edited summary of an event across media; Event Film does not remain a second type with the same meaning.
Why: Joe said, “all reccomendations in event recap approved”. A recap selects highlights, experiences or takeaways. A complete keynote, interview or single event photograph retains its own form; an event gathers those related pieces without making them all recaps. Keep the source wording and evidenced event relationships.
This accepts the meaning and name on the review page. Browsing placement remains separate. Existing Event Film and Event Coverage assignments require individual checks before relabeling; the answer does not convert records or certify event navigation, classification or pipeline implementation.
Last reviewed 5 October 2026.
Ruled on finding and comparing product features (5 October 2026)
A feature can appear in several supported categories while keeping one identity.
Why: Each category membership needs support. Different research entrances should not create duplicate features or inflate comparison counts.
Compare specific, defined capabilities where evidence supports them, preserving the original named feature or bundle.
Why: A useful comparison may concern only part of a broader named feature. Keep the company’s wording and distinguish a documented claim from an observed demonstration. Do not split every feature into tiny operations.
Joe approved both recommendations on the feature review. Exact category names, storage, identity resolution and complete pipeline behavior remain to verify. The separate graphics review remains open for a deeper page-type and section audit.
Last reviewed 5 October 2026.
Ruled on hands-on interaction (5 October 2026)
Hands-on Interaction includes deliberate handling and adjustment with the hands, as well as operating controls. A posed hand merely holding a product is insufficient. Describe the action and keep whether the product is actually being used for a task separate. Use can happen without touch, and handling does not always establish use. Supported descriptions may overlap.
Joe accepted the explicit recommendation in Using a product and handling it. This settles its one meaning choice. It does not merge older labels, narrow the complete Product in Use shelf or certify existing assignments.
Last reviewed 5 October 2026.
Ruled on imagery, sections and intended use (5 October 2026)
Find email imagery through evidenced email context and section roles; do not add Email Header as a maintained type or intended-use filter now.
Find article imagery through its evidenced article relationship and placement; do not add Blog Header Image as a maintained type or intended-use filter now.
Keep social posts, captions and ordered attachments connected; do not add Social Post Graphic as a maintained type or intended-use filter now.
Digital Wallpaper is a searchable intended use for artwork explicitly prepared or offered as a device background, not a whole-piece type.
Why: Joe accepted the current recommendations after the website, page and section review. An asset keeps its own form while its context explains where it appeared or what it was prepared for. A header or hero can contain image, video or text; position alone is insufficient evidence. A publisher offering a wallpaper establishes intended use, not installation.
The answered review retains the examples and alternatives. Earlier proposals to maintain the first three intended-use filters are superseded. Complete section vocabulary, storage, prompts, existing assignments and end-to-end research behavior remain audit and implementation work.
Last reviewed 5 October 2026.