Anatomy of a video that holds.
A hook earns the first two seconds. Structure earns the next ninety. Every video that performs — yours included — follows the same five beats, then gets cut into six sizes from one shoot. Here's the blueprint.
Hook → Stakes → Payload → Proof → CTA
Memorise this order. Every strong video you have already follows it; every weak one skips a beat.
The curiosity gap (see the Hooks masterclass). Spoken + on screen, frame one.
Why it matters to them, now. Name the problem: "you'll waste money," "you'll quit frustrated," "you bought the wrong one."
Deliver on the hook. One idea, taught clearly. Rajesh explaining, product in hand. Resist cramming three points into thirty seconds.
Show, don't tell. The swatch, the before/after, the finished piece, the shelf. Drop a credibility line ("10,000+ artists, all genuine").
One ask. "Shop the kit," "Subscribe," "Visit the store." Said out loud, shown on screen, and on an end card.
Same beats, different runtimes
| Beat | 6s bumper | 15s | 30s | 60–90s |
|---|---|---|---|---|
| Hook | 0–2s | 0–2s | 0–2s | 0–2s |
| Stakes | — | 2–4s | 2–6s | 2–8s |
| Payload | 2–5s (one claim) | 4–11s (one point) | 6–22s (two points) | 8–70s (the lesson) |
| Proof | woven | woven | 18–26s | woven + a hero shot |
| CTA | 5–6s | 11–15s | 26–30s | last 5–8s + end card |
The shorter the cut, the fewer points. A 6-second bumper makes one claim. A 30-second ad makes two. Never try to say everything — say one thing well.
The retention curve is a cliff. Build ledges.
Viewers leak away every second. Your longer videos (the 89-second "10 Reasons," the 156-second Acrylic Range) lose people because they're one static shot with no reason to stay past the point they've "got it." Three tools fix this:
Re-hooks
Every ~10–15s, a new mini-promise: "but here's the part nobody tells you…"
Visual variety
Cut every 2–4 seconds. Talking head → macro b-roll → product → result. Never hold one frame for 30s.
Escalation
For listicles, save the best reason for last and tease it early: "and number 1 will surprise you."
Re-cutting "10 Reasons to Visit"
29,400 views, founder energy — and still leaves a lot on the table. Here's its actual filmstrip, then the re-cut.
What it does now
Opens mid-sentence as the "10 Reasons" card slowly assembles. One locked-off angle for 89 seconds. All ten reasons at equal weight. No store b-roll. CTA is soft.
The re-cut
Open on Rajesh saying the single most surprising reason ("Reason #1: we'll mix any colour you bring us"). Then "…and 9 more." Cut in 2–3s of real store b-roll between reasons. End on one hard CTA: "Come see for yourself — Kala Ghoda."
Film once. Get six.
This is the efficiency multiplier. Every Rajesh shoot should leave the edit bay as all of these:
| Asset | What goes in it | Where it runs |
|---|---|---|
| 6s bumper | The single sharpest claim + logo | Cheap reach / reinforcement |
| 15s | Hook + one point + CTA | In-feed, Shorts |
| 30s | Hook + two points + proof + CTA | Skippable in-stream (the workhorse) |
| 60–90s | The full lesson | Channel + in-feed |
| 1–3 Shorts | The most satisfying moments, vertical | Shorts feed (organic + paid) |
| Wide + tall | Both crops from one wide shoot | In-stream/TV + Shorts |
One hour with Rajesh → 4–6 episodes → ~30 ad-ready assets. That's why the founder format isn't just better creative, it's cheaper creative.