The team at the neighborhood bakery needs to promote its weekend pastry box, but the folder contains notes and no coherent asset set. The work must make sense to busy parents on a phone and lead naturally toward the choice to place a preorder. The producer organizes the assignment around approval discipline and uses a costly production mistake as the first point of recognition.

A weak route would mention a costly production mistake repeatedly without showing its effect on a real choice. The producer writes a rejection list before drafting: invented features and unsupported performance claims. Those risks shape the claim, composition, shot list, and approval checks.

The working brief fits on one page: the neighborhood bakery is promoting the weekend pastry box to busy parents, and the single communication goal is to place a preorder. Tone is warm and unhurried. Required inputs are approved wording, timing, cost, eligibility, delivery method, and contact path; unsupported results and false urgency are prohibited. Constraints specify spelling, mandatory and avoided terms, source permissions, the 1:1 ratio, safe zones, 28 seconds, file formats, responsible people, and release time. Approved facts are separated from questions and examples so assumptions cannot become public claims.

The planning sheet translates that search language into its underlying job: to evaluate unfamiliar image-making search language without implying endorsement. The acceptance test is a controlled test against one deliverable; speed alone is not an outcome. The phrase guides the job, but the weekend pastry box stays as the one illustrative case.

A vertical script capped at 28 seconds shows one decision, from loose notes about the weekend pastry box to a reviewed asset. For three seconds, voiceover names a costly production mistake, screen copy stays under five words, and the camera shows real planning notes after private details are removed. Shot two collects inputs for four seconds; shot three shows the illustrative choice for six; shot four pauses on a human correction for five; shot five uses the remaining time for the approved next step. The shot sheet separates voiceover, on-screen text, direction, duration, source or assumption, and review note into six columns.

The approved route expands into one explanation, a caption, a two-line hook, a five-card carousel, a voiceover, and three headline options. Each card has one job: identify friction for busy parents, list inputs, demonstrate a hypothetical decision, review it, then invite people to place a preorder. If a fact is absent, the draft carries an editor query and stops short of publication wording. A line stays only if it provides a method, example, criterion, or risk that changes the work.

The visual is not decoration; it must place the product inside a believable daily routine within the weekend pastry box story. Its prompt names the weekend pastry box, places the decision in the foreground, keeps support behind it, limits color to navy, pale blue, and restrained yellow, and requests 1:1. No essential detail enters the mobile crop or caption-safe margins; real copy is added later as an editable layer. 'Room to Begin' and 'Inside the Process' are hypothetical options for internal layout testing, not claims of availability or proof.

Each channel receives a fresh edit of the approved argument rather than a resized duplicate. The written channel preserves the decision process, the image channel leads with the clearest frame, and the carousel reveals one step at a time. The vertical clip puts a costly production mistake in the first two seconds with large captions; a longer video retains the example and source notes; a community post asks one focused question before inviting people to place a preorder.

The schedule also reserves enough time for correction, source checks, and another careful final reading before any finished asset is queued. The minimum viable set for the weekend pastry box contains a master argument, a flexible visual frame, related caption copy, and a timed demonstration of the same decision. The owner reviews that chain before variants are made, because a weak assumption copied into twelve exports becomes harder to locate and more expensive to repair. The production board records the asset, audience purpose, supporting fact, destination format, approval state, reviewer, and how it helps people place a preorder. Measurement uses a preselected behavioral signal rather than a vague impression; the owner records the publishing window and reach before interpreting inquiries, saves, or replies. The result informs the next approval discipline decision but is not presented as universal proof, a guaranteed lift, or a customer testimonial. Version control is deliberately plain: drafts receive descriptive names, approval is written down, and obsolete exports move out of the active folder so nobody publishes an attractive but rejected file. Each exported derivative is opened in context before release, catching platform compression, unexpected cropping, unsafe captions, broken links, or a reordered message that the working file did not reveal.

The fact editor traces statements to approved inputs and dates any official-rule lookup; the copy editor checks audience, tone, spelling, case, and prohibited implications. The visual reviewer checks 1:1 derivatives, mobile crop, try-nanobanana contrast, hierarchy, spelling, doubled characters, icons, anatomy, edges, shadows, duplicate objects, interface structure, and accidental marks. A cold reviewer watches for changing objects, warped text, caption errors, unsafe margins, mispronunciation, uneven pace or audio, weak boundary frames, and lost meaning without sound. Final approval is recorded when the formats hold the same facts, maintain the stated warm and unhurried tone, and make the next action appropriate for busy parents.

Specific model failures in this case include invented features and unsupported performance claims. Other risks include continuity errors, offensive ambiguity, accidental imitation, warped lettering, strange anatomy, and fabricated case-study claims. A second reviewer compares each output with the source, inspects artwork at full size, and reads the script without prompt history.

The final folder separates the source sheet, approved route, editable copy, layered artwork, caption file, shot list, provenance notes, and export checklist. With one source of truth, the producer can change a detail and identify every affected placement. The neighborhood bakery finishes with connected assets that can be checked and changed without pretending automation supplied judgment.