Timeline — sources

Reference ad (Foreplay #7, analysed 10-06 in ~/research/davidaistar/foreplay/07_timeline.gemini.md)

IM8 Health, Daily Ultimate Essentials, 1:51, 9:16, ~80% AI. Paraphrased beat by beat: - 0:00-0:10 hook: a man pulls a beer from the fridge and lies on the couch, the product sachet beside him; VO sets up "if a regular drinker took this every morning for three weeks, here's what happens". Relatable vice + one small habit + a curiosity gap. - 0:10-0:38 first days: he walks upstairs with the dog, glowing 3D gut and body graphics, ingredient macros; VO maps each ingredient to a symptom (gut settles, liver strain eases). - 0:38-0:51 calendar flips to Week 1: 3D stomach/skin graphics, he touches his face; VO: digestion, skin, lighter mornings. - 0:52-1:07 Week 2: glowing brain, ingredient macros, he pours the alcohol away and runs outside; VO: fog lifts, stamina without coffee. - 1:07-1:27 Week 3: he wakes refreshed, plays with family, lifts weights; VO: night wake-ups stop, leaner, present with family. - 1:27-1:42 proof: grid of real creators drinking it, real product footage; VO: customer count, ingredient count, celebrity co-founder, certification. - 1:42-1:51 end card (optional, --end-card; off by default per Fish 10-06): celebrity faces + 30-day money-back guarantee. - Production tells: morphing glasses, warped fingers, background warping; generic corporate music; real footage only at the end.

Keep: the "if X did Y for N weeks" promise as one hook variant; time moving forward in visible jumps; one symptom retired per stage; the guarantee as the last word. Change: IM8 is a character-light explainer over 3D anatomy; on Dawn that pattern got 0 purchases (MAP.md "Don't build"), so ours is ONE family with mom narrating and at most one anatomy-style insert. IM8 maps ingredients to symptoms; we map stages to her objections (wear → wake → lasts → her morning). No UGC grid (we have no real UGC and the Okendo set is imported); the end card (optional, --end-card; off by default per Fish 10-06) carries the real product cutout + guarantee. No celebrity, no certification we don't hold.

davidaistar statics notes

• analysis/gemini_HEa4AVv-ths.md + analysis/guide_copy_images_shorts.md H2: the Timeline is a middle-of-funnel template ("product shot paired with Day 1, Day 2, Day 3 progressive benefits"); job = kill the "how long until this works" objection; claimed AOV lift via bulk buying (unverified). Generation method: improvise first, then an edit pass for exact text; specificity beats generic benefit lines.
• analysis/diff_statics.md: top-5 change #1 = build the Timeline (day1/day3/day14) and iMessage templates; "No Timeline template ... anywhere in the 20-template library". Our 20 templates confirm it; C01 (Lunisk progression) is the nearest grid and bans time labels.
• REPORT.md: format table row 7 "Timeline (IM8 ...) YES animated, cartoon-h3, Medium. Must carry a character; no-character explainers got 0 purchases"; §11 maps his templates to recipe-B statics (Dawn = problem-aware, so the MOF set).
• Keep: the stacked progression, three stages, the objection it answers. Change: no day numbers (qualifier ban; see beat-sheet.md §3), no blank-prompt improvise without the locks (his rule caused our 9-03 palette drift); gen_statics' 10-06 --improvise keeps PROD + BRAND LOCK on pass 1 and is the route once it can read waivers (SKILL §9), product = our real cutout, never redrawn.

Our receipts

• oneweek song (sa-ontime-04-relatab-animation-song-oneweek-a-v2copy): $1,106.27 spend, 26 purchases ($42.55 CPA), ROAS 1.81 (output/dawn-combo-map-2026-10-05/songs_all.json). Its lyrics already run day by day (Monday...Thursday, Saturday, then Monday) inside an experiment week (songs_scripts.json). Exhibit A: $292.33 / 9 purchases ($32.48), ROAS 2.48.
• MAP.md: stake must land on her and be caused by the mornings; character-less explainer VO = 0 purchases; testimonial statics are the five weakest pairings; static payoff = independence/calm.
• Owned survey 9-15 (research/customer-evidence-2026-09-15.md): wake doubt 82/156 (52.6%), wear doubt 29/156 (18.6%), waking them myself 142/156, phone alarms 124/156; S01-S08 verbatims.
• Category evidence (research/voc-competitor-reviews-2026-10-02.jsonl): q1311, q1358, q1372, q1382, q1404, q1435 (positive progressions), q1448 (first-night failure).
• research/bottleneck-diagnosis-2026-10-03.md: no post-use measurement; Okendo imported; proof deficit is the #1 bottleneck.
• Template-driven static render path and waivers: output/dawn-iterate-2026-10-04/statics/ (batch.py, build_prompts.py, render.py, check_gate.py, RENDER-LOG.md: worn bands must be real-photo composites; product standing on its edge failed review).
• Clay reference sheets ready for reuse: output/cartoon-h3/_refs/dawn-songs-2026-10-04/clay/.

starpop.ai articles (written versions, read 10-06 from ~/research/davidaistar/starpop/txt/)

• ai-static-ads-that-convert, starpop-static-ads-guide: the timeline static is his mid-funnel template (Day 1 / Week 2 / Month 1, kills the "when will it work" objection, claimed AOV lift); "From [pain] to [outcome]" gets stronger with a timeframe; the Pain-Point-Gone template adds a "without [objection]" clause. Adopted: the "without …" clause on stage 3 (SKILL §5b). Not adopted: dated stage labels and "in 2 weeks" timeframes, because the statics qualifier ban and our missing post-use data (no day-14 survey) make any dated promise unbacked (beat-sheet truth table); the AOV claim stays unverified (a band isn't a refill).
• create-and-clone-competitor-ads-with-ai: keep the visuals honest to the script (a frustrated character is never shown already holding the result). Adopted as the §7 rule + §10 check, because a timeline lives on stage order.