Six Shorts, 25.9 Million Views: Each One Put Its Whole Setup on Screen in 7 Seconds or Less. Copy the Cold Open Sheet
# Six Shorts, 25.9 Million Views: Each One Put Its Whole Setup on Screen in 7 Seconds or Less. Copy the Cold Open Sheet
We ran ClipNovia video analysis on six YouTube Shorts that each had more than 2.1 million views on October 28, 2026 (25.9 million combined). The analyzer put the hook at 1 to 7 seconds in every one (median 2.5s), and in each case the first seconds already showed the obstacle, contrast or stake that the last beat pays off. The hook got longer as the clip got longer: the two 11 to 12 second clips used 1 to 2 seconds, the two clips over 50 seconds used 4 to 7.
**In this sample, nobody opened with a hook line and figured out the story later. The first frame was the problem, and the last frame was the answer.**
This is six videos from one feed on one day. Treat it as a teardown, not a law. The limits are listed near the bottom, and the template you can copy is in the middle.
## What we pulled and how
On 2026-10-28 we queried ClipNovia Discover for YouTube Shorts above 500,000 views (16 results). The day before, [our AI animation teardown](/blog/ai-animation-shorts-teardown-flipped-perspective-skit) covered seven cartoon-style clips from the same feed and set aside the brainrot, game and meme clips. This post covers six of those other clips, plus one faith-based story that the cartoon batch did not include. We ran ClipNovia video analysis (standard mode, medium effort) on each. View counts, lengths and post dates come from the feed. Hook type, hook length, cut rhythm and story beats come from the analyzer output. We did not watch each clip frame by frame ourselves.
| # | Short (channel) | Views | Length | Posted | Analyzer hook type | Hook length | Avg cut | Progression |
|---|---|---|---|---|---|---|---|---|
| 1 | [Brainrot ahh textbook](https://www.youtube.com/shorts/yEVCm6ibRYI) (Cyall) | 5.3M | 12s | 2026-10-08 | humor open | 2s | 2.2s | escalation |
| 2 | [Krishna Entered the Waters of Kaliya](https://www.youtube.com/shorts/Q1YJ-kUJ4BE) (MAHIANX) | 7.6M | 51s | 2026-09-23 | dramatic visual reveal | 7s | 7.3s | escalation |
| 3 | [The Evolution of the Verity Song Is INSANE](https://www.youtube.com/shorts/9yVyRJt5C8A) (NO FACE) | 4.3M | 32s | 2026-08-19 | claim / tease | 2s | 3.9s | escalation |
| 4 | [RUN for 67 Meme Story](https://www.youtube.com/shorts/kJ-bvpXFTaA) (Sosonya) | 3.7M | 36s | 2026-08-23 | dramatic visual reveal | 3s | 3.2s | escalation |
| 5 | [Steal an Egg Made a BIG MISTAKE](https://www.youtube.com/shorts/1KR8zLgkqUg) (Louquit) | 2.9M | 55s | 2026-10-03 | dramatic visual reveal | 4s | 5.3s | none |
| 6 | [Jesus vs. Evil](https://www.youtube.com/shorts/f7lyuxcGlcs) (Jesus Daily Short) | 2.1M | 11s | 2026-08-30 | dramatic visual reveal | 1s | 3.3s | escalation |
Views are a one-day snapshot of videos posted between 2 and 10 weeks apart, so they are not directly comparable across rows.
## Finding 1: the first frame states the problem
Using the analyzer's description of each opening:
- The textbook clip opens on a workbook page with an absurd illustration, then flips page after page.
- The 67 clip opens on a comically steep ramp with four racers at the bottom and a prize at the top.
- The Steal an Egg clip opens by shattering a game icon while the narrator says the game was almost deleted forever "for a ridiculous reason".
- The Jesus vs. Evil clip opens with two objects held at the top of a staircase and a spoken "Ready, Go!".
- The Krishna clip opens on a poisoned river with villagers trembling.
In each case the viewer can tell what is at stake before any dialogue explains it, and the answer is withheld until the final beat. That is a different job from a generic hook line, which is what most "25 hooks" lists give you (for example [Metricool's viral video hooks list](https://metricool.com/viral-video-hooks/) and [CapCut's short-form hook formulas](https://www.capcut.com/create/short-form-video-hooks), judged here from their titles only).
## Finding 2: hook length scaled with runtime
| Runtime | Clips | Analyzer hook length |
|---|---|---|
| 11 to 12s | Textbook, Jesus vs. Evil | 2s, 1s |
| 32 to 36s | Verity, 67 | 2s, 3s |
| 51 to 55s | Krishna, Steal an Egg | 7s, 4s |
The advice to nail everything in the first three seconds is not wrong, but in this sample the longer stories used a longer opening to build the stake, and still held 2.9M to 7.6M views. A 7 second establishing shot worked for a 51 second clip. It would eat two thirds of an 11 second one.
## Finding 3: in the longer clips the turn lands at 56% to 69% of the runtime
Taking the analyzer's timestamp for the moment the story flips:
- RUN for 67: the underdog starts sawing the support under the finish platform at about 0:25 of 36s (69%).
- Steal an Egg: the platform crackdown starts at about 0:30 of 54s (56%).
- Krishna: the dominance reversal begins around 0:33 of 51s (65%).
Taken together, the turns begin at 56% to 69% of runtime. The short clips have no such mid-point turn: the textbook and the staircase drop are both single-escalation runs that end on their punchline. If you are making something over 30 seconds, plan one reversal in the 55% to 70% band, not at the end.
## Finding 4: fast cuts still did not separate them
Average cut length by views: the textbook clip cut every 2.2s and had 5.3M; the Krishna clip cut every 7.3s and had the most views at 7.6M. The three fastest cutters (2.2s, 3.2s, 3.3s) totalled 11.1M; the three slowest (3.9s, 5.3s, 7.3s) totalled 14.8M. This matches what we found in the [animation teardown](/blog/ai-animation-shorts-teardown-flipped-perspective-skit), where cut speed also did not separate the winners. Cut speed is a style choice here, not a lever with visible effect.
## Finding 5: five of six escalated, and almost none closed with a request
Five of six were labeled escalation by the analyzer. Only one (the Verity song timeline) ended with a direct audience question, asking which version was best. The Steal an Egg clip placed its subscribe call at the story's low point, around 0:37 to 0:44, not at the end. One example each is not a pattern, so we do not recommend either as a rule. We only note that a closing request was not universal in this sample.
## The Cold Open Sheet (copy and use)
Fill it in before you open an editor or a video model. If a line is blank, the clip is not ready to generate.
```
COLD OPEN SHEET
Runtime: ___ s Hook budget: 1-2s (up to 15s) | 2-3s (30-40s) | 4-7s (50s+)
1. FIRST FRAME (must work with sound off)
The obstacle or contrast the viewer can read at a glance:
"A ______ that looks ______ because ______"
2. STAKE LINE (8 words or fewer, spoken or captioned)
"______ almost ______ for a ridiculous reason"
3. WITHHELD ITEM
What the viewer waits for. Do not show it before the last 20%:
______
4. ESCALATION STEPS (one new thing every 3-4 seconds)
a) ______ b) ______ c) ______
5. THE TURN (clips over 30s only, place at 55%-70% of runtime)
The moment the rules change (shortcut, reveal, reversal): ______
6. LAST BEAT
The payoff in one frame: ______
Optional comment question: ______
```
Worked example (our own, not from any of the six): runtime 30s, hook budget 2-3s.
1. First frame: three paper boats at the edge of a bathtub, the finish line is a rubber duck at the far end.
2. Stake line: "Two boats paddled. One pulled the plug."
3. Withheld item: how the third boat wins.
4. Escalation: a) boat one tips over, b) boat two stalls on a soap wave, c) boat two overtakes.
5. The turn at about 0:18: boat three drains the tub and slides to the duck.
6. Last beat: the duck at the drain, boat one floating upside down. Question: "Cheating or genius?"
## Where this sample is weak
- Six videos, picked from the feed by us. We skipped some clips, so this is not a random sample.
- Correlation only. We do not know why any of these videos got its views, and two of them (7.6M and 2.9M) are far from the rest.
- Hook type, hook length, cut rhythm and timestamps are analyzer output, not hand-timed.
- Views are a snapshot from one day. Posts range from 2 to 10 weeks old.
- Four of the six lean on someone else's characters or worlds per the analyzer descriptions (Minecraft, Roblox, Brawl Stars characters, a Friday Night Funkin mod), and two use faith-based figures. We did not see any of the six described as built on fully original characters, so copy their structure, not their assets. Our post on [whether you can legally recreate a viral video with AI](/blog/is-it-legal-to-recreate-a-viral-video-with-ai) covers the legal side, and the [AI disclosure decision card](/blog/youtube-ai-disclosure-decision-card-faceless-shorts) covers labeling.
- We did not generate a video from the sheet, so we cannot say it improves performance. It is a planning template built from what these six did.
## How to use it with ClipNovia
Paste any Short URL into ClipNovia's video analysis and you get the same fields we used here: hook technique and length, structural blueprint, cut rhythm and retention mechanics. Run it on three Shorts in your niche, fill in the Cold Open Sheet for your own idea, then use the blueprint timestamps to check your plan against theirs. You can [try the analysis here](https://clipnovia.io/?utm_source=blog&utm_medium=post&utm_campaign=short-form-cold-open-sheet-six-viral-shorts).
For more on method, see [how to reverse engineer viral videos](/blog/how-to-reverse-engineer-viral-videos), and for hook wording once the structure is set, [viral hook formulas for short-form video](/blog/viral-hook-formulas-for-short-form-video).
## Questions we can answer from this sample
**How long should a Short hook be?** In these six, the hook ran 1 to 7 seconds by the analyzer's measure, and scaled with runtime: 1 to 2 seconds for clips near 11 seconds, 4 to 7 seconds for clips over 50 seconds.
**Do fast cuts help?** Not visibly here. Average cuts ranged from 2.2s to 7.3s, and the slowest cutter had the most views.
**Where should the twist go in a 30 to 55 second Short?** In the three longer clips here it began at 56% to 69% of the runtime.
_Data pulled 2026-10-28 from ClipNovia Discover. Recheck by 2026-11-28; view counts will have moved by then._
Related: [the same feed sorted by length instead of structure](/blog/youtube-shorts-length-feed-data-30-seconds-vs-15)
*All three teardowns are summarized with a copyable five-field line in the [reverse engineering guide](/blog/how-to-reverse-engineer-viral-videos).*
Related: [Where the 85% sound-off caption stat comes from, and a caption test to run](/blog/sound-off-statistic-85-percent-origin-captions-test)