Reels Script

Reels Script

Reels are Instagram's discovery engine: most Reel reach comes from non-followers, decided by how viewers respond — not by follower count. But Reels are not TikToks with a different logo:

1. The hook window is ~3 seconds — they decide whether a Reel is pushed wider or throttled. Layer it: verbal + on-screen text + visual.
2. Watch-through and sends rule. Watch time is the top signal; sends-per-reach (DM shares) are weighted roughly 3–5× more than likes for reaching new people. Script for both.
3. Write the muted version first. Most Reels viewers watch on mute — if the on-screen text doesn't carry the story, the script fails before the audio matters.
4. Eligibility gates come before ranking: original, no TikTok/CapCut watermarks, has audio, under the length cap, clear niche. A watermarked reupload is throttled at the gate.
5. The Reel also lives on your grid — plan the cover frame; it can't be edited after publish.

(Full mechanics in references/reels-mechanics.md.)

Step 0 — Read the foundation first

Load brand-profile.md and voice.md. Voice carries into the spoken script, on-screen text, and caption. Carry guardrails and compliance rules.

Step 1 — Lock the concept, goal, and length tier

• One idea — a Reel does one thing.
• Goal — reach, sends, saves, follows, or comments (shapes the close/CTA).
• Length tier — ~15–35s is the reach sweet spot; up to ~90s for tutorials/storytelling that earns it; up to ~3 min stays recommendation-eligible. Let the idea pick the length; never pad. See references/reels-mechanics.md.

Step 2 — Decide the audio (and write sound-off anyway)

Original audio (your voice) for most founder/brand content; a trending audio where it genuinely fits — but Instagram's licensed library is native-only: the script names the audio direction, the human adds the track in-app (WoopSocial can't attach trending audio). Either way, the Reel must work on mute. Live trend research → trend-jacking. See references/formats-and-audio.md.

Step 3 — Write the 3-second hook

Use hook-writer. Fire on all three channels at once — verbal + bold on-screen text + a striking first frame — with no intro, no logo, no "hey guys." The on-screen text hook must work silently, and the hook must be true: a bait-and-switch creates a mid-video retention cliff the algorithm punishes. See references/hooks-and-retention.md.

Step 4 — Structure for watch-through and sends

Front-load value, open a loop early, change the visual every ~2–4 seconds, cut all dead air, and engineer a loop or rewatch trigger so the ending feeds the start. Then ask the sends question: who would DM this to whom, and why? Relatable, useful-reference, or conversation-starting beats travel; generic tips don't. See references/hooks-and-retention.md.

Step 5 — Write the production doc + cover brief

Output a shootable doc — a row per beat with shot · on-screen text · spoken · audio · timing — plus a one-line cover-frame brief (which frame, or a custom cover with bold text in the grid-safe center). Keep hook text, faces, and CTAs inside the safe zone — exact zones live in instagram-reels-publishing.

Step 6 — Instagram SEO + caption + hashtags

Instagram search is keyword-based now: put the target keyword in the first ~3 seconds spoken, the on-screen text, and the caption's first sentence — and the visual must match the caption. Keyword research + the search-first caption → instagram-seo; caption craft → caption-writer; 3–5 topic-label hashtags, not thirty.

Step 7 — Test with Trial Reels; produce film-first or with the Veo pack

For unproven hooks/formats, recommend a Trial Reel (shown to non-followers first — a free test that doesn't touch follower-facing stats). Default to native filming for faces and voices. For faceless niches, b-roll, or unfilmable shots, use the bundled Veo prompt pack (references/veo-prompt-pack.md; API layer in tools/integrations/veo.md) — and follow Meta's AI-labeling rules for generated media (disclosure on Instagram is native/manual, not automatic).