Warehouse — sources

Evidence status: UNGROUNDED (2026-10-06)

• Fish listed "AI warehouse ads" as format #11. ~/research/davidaistar/foreplay/links.txt jumps from 10_character_arguments to 12_musical; REPORT.md: "13 unique Foreplay ads (#10 and #11 are the same link)".
• Checked foreplay/10_character_arguments.gemini.md: Penrose Skin scented lotion, 62 s, 3D lotion-jar man vs cologne-bottle man podcast debate. Not a warehouse ad; it belongs to format-courtroom.
• No warehouse reference has been watched or dissected. Everything format-specific in this skill is built from the format as it commonly runs (staff or founder among stocked shelves, packing orders, "we can't keep these in stock", "here's why N parents ordered this", scarcity/demand proof, behind-the-scenes authenticity), bound to Dawn doctrine and our own data.

Closest analysed ad

• 05_founder — Azure Batana Oil, 3:10, 9:16 (foreplay/05_founder.gemini.md). Paraphrased shape: free-offer pattern interrupt → insider credibility (years in the industry) → agitation over fast b-roll → personal anchor (his mother) → mechanism (penetrates vs coats) → brand origin → proof (the mother's results) → BOGO + 90-day guarantee. Almost all AI: spokesperson + lip-sync + animated b-roll; tells: glossy skin, rigid torso, morphing background text.
• Keep: fast b-roll carrying the argument; mechanism before offer; guarantee as the close.
• Change: no talking spokesperson in the default (borrowed-person authority underperforms in recipe B; lip-sync tells); 30-45 s not 3 min; proof = demand + the visible silent buzz, not a relative's results; no "free" offer (the live offer is the PDP's).

davidaistar method notes that apply

• analysis/diff_ai_ugc.md: product-in-hand via omni-reference scored 3-4/10 with fused fingers on our band (MiniMax H3, 10-04). Rule here: tier-A hand/product shots come from real footage or composites; Seedance 2.5 gets a one-unit proof before any batch. His face-blur trick to pass likeness filters is not adopted (never work around a provider filter).
• analysis/diff_cloning.md: dissect any source; a lane-leader competitor's dissect is a gap map, never a template; the brief must fail the competitor-swap test. The dissector (scripts/ad_dissector.py) is the grounding tool.

Our receipts

• MAP.md §2B recipe B: brand voice $201K at 1.87; authority guarantee-only 1.84, science 1.80, borrowed-person 1.50; testimonial pairings the 5 weakest (1.21-1.31).
• aggregate.txt: statics authority social-proof $17,469, ROAS 1.83 (n5, 4 W / 1 L); songs social-proof $182, 0.72 (n1); angle "proof" $468 (n1); angle "witness" $301, 1.04 (n1).
• feedback_guarantee_positioning: guarantee-led headlines died; guarantee = risk-reversal chrome; HiStrips saturates "tested on 10,000+".
• reference_dawnbands_packaging: retail pack is a stand-up ziplock pouch; 3D pouch PNG for composites; Higgsfield element @dawn-packaging; the packaging opt-in trap in char-swap prompts.
• project_dawnbands_broll_seedance_playbook: fidelity block (diagonal ribbing, solid black inside and out, no module, no glow), product-alone buzz renders as a rattle on a hard surface, physics anchors, reaction-primary wake shots, sleepwear on any in-bed human.
• project_vo_broll_lane + skills/vo-broll/references/production-handoff.md: voice-clocked cuts, one cast per ad, end card (optional, --end-card; off by default per Fish 10-06) ≤2 s, Gemini scores real footage 4/10 on rubric artifacts (report, don't adjudicate), runner cache/shared-module gaps.
• Live PDP JSON 2026-10-06 (read-only): wake-up-band-adhd, 4 variants all available at $44.99 (price test since 10-04). So today no scarcity line is true.
• Clip library 2026-10-06: 249 clips; no warehouse/packing-room footage; pouch appears only inside Lori finished ads (tagged finished-ad, excluded by the runner unless pinned to a clean window); c2b-box-open-palm is a white-box unboxing (wrong packaging).

starpop articles (creator's own write-ups, read 2026-10-06; ~/research/davidaistar/starpop/txt/)

• Applied: facebook-instagram-ad-sizes-and-safe-zones + how-to-preview-a-facebook-ad-before-publishing → §10 placement gate (the article names the UI regions but gives no percentages; the 14/35/6 % Reels box is Meta's published spec, cross-checked 10-06). are-ai-generated-ads-allowed-on-meta → §11 upload hygiene (Advantage+ enhancements off, live preview day 1). meta-video-ad-metrics-hook-rate-hold-rate-ctr-cpa, how-much-budget-to-test-meta-ad-creatives → §11 read/floor lines where this format needed them.
• Read, nothing new for this format: ai-ad-formats-guide (format ↔ job table, shot-list order, references before video: already core §4), fully-ai-generated-ads-are-here-to-stay (strategy essay), how-to-calculate-break-even-roas-for-meta-ads (proposed for core §6, not format-specific), how-to-add-correct-captions-in-any-language + how-to-fix-wrong-capcut-auto-captions (script-as-source captions = our assemble rule already; his SRT → CapCut/Premiere import maps to Remotion per core §2a).
• Not adopted: (1) "an AI actor must not voice a fictional first-person experience / fabricated testimonial" (formats guide, AI-ads article, organic-vs-paid) — conflicts with Fish's claims unlock (10-02) and our dramatized mom-narrator recipe; risk is Fish's call, recorded here only. (2) His 25-char vertical caption line cap — VideoAd chunks by its own MAX_PHRASE_CHARS 28 + two-line fit, Fish-set. (3) Organic-first discovery (ai-organic-social-vs-paid-ads) — Dawn's message and checkout path are proven, Testing CBO is our discovery lane. (4) His model picks (Kling O3 Pro, Sora 2, P-Video) — registry roles only; untested candidates.