The View Is Weighed At 4 Checkpoints

The platform stopped counting views a while ago. It weighs behaviour inside the view, and it does that at 4 checkpoints, each 1 tracked on its own. 1 second: did anyone stay past the first frame. 3 seconds: did they commit to the premise. The middle: where the dip shows up. The end: was the piece finished, rewatched or shared. Because the 4 are read separately, a piece can open strongly and still die. A strong first second buys nothing at the 3 second mark if the premise is still unclear, and a strong middle cannot repair a hook that pulled the wrong room. The practical consequence is 1 line. There is no single number called performance. There are 4, and each 1 has a different owner inside the edit. So the method is to read the 4 in order and repair the first 1 that is failing. Working on checkpoint 3 while checkpoint 1 is losing people is the most common way a team spends 3 weeks on the wrong end of a 7 second film.

The 1 second mark is won by the visual. It registers before audio does, and a large share of viewers watch the opening moments muted, with captions carrying the words. The average mobile attention decision takes about 1.7 seconds. That is the entire window, and it is shorter than most people think when they are staring at a timeline. So frame 1 carries movement, contrast, a bold text overlay, or a face with energy. Not a title card, not a slow push in, not a logo. The opening matters more than the length of it. 63 per cent of top-CTR videos hook inside 3 seconds, and the first of the 4 checkpoints is already decided well before that - the decision is made in the first frames, and everything after it is a consequence. 1 test before posting. Pause on frame 1 and ask whether a stranger who has never heard of this could tell, without sound, that something is happening.

The 3 second checkpoint is won by the hook, and the hook has 2 jobs rather than 1. It has to stop the scroll, and it has to signal the niche so the platform routes the piece to the right viewers. That second job is the 1 most teams skip. A confusing hook earns views from the wrong audience, the wrong audience leaves early, and the retention that the feed reads collapses. The piece dies with a healthy view count and a broken curve. The hook stack runs in a fixed order across the first 3 seconds. Visual trigger at 0.0 to 0.5 seconds: movement, contrast, bold text, a face with energy. Verbal trigger at 0.5 to 1.5 seconds: 1 clear sentence carrying a promise, a curiosity gap or a problem. Context lock at 1.5 to 3.0 seconds: who this is for and what they get. 7 frameworks earn their place, and each 1 is a shape rather than a line. Shock to logic. Specific result. Open loop. Hot take. Point of view pain. Speed-run tutorial. Proof first. The killers are just as fixed. Opening with a greeting. A generic promise. A slow setup. No on-screen text for viewers watching muted. Any 1 of those 4 spends the 3 second checkpoint before the content has begun.

The middle is not a checkpoint that can be won once. It is the section where the curve reveals itself, and the dip is almost always the same shape: a promise made in the hook that the delivery did not keep. The bands make the reading simple. 70 per cent or better intro retention is strong. Under 40 per cent is dead on arrival. Top creators sit between 60 and 80 per cent overall, and the sharper tell sits at 75 to 85 per cent at the 30 second mark. A video does not stall at about 300 views because the content is bad. It stalls because nobody stayed past the hook, and the middle is simply where that gets recorded. 2 mechanisms hold the middle. Pattern interrupts every 5 to 10 seconds, and a cut rhythm that keeps moving - cuts every 0.5 to 2 seconds with most landing at 1. The benchmark from the strongest openings is about 1 shot change every 1.77 seconds, roughly 19 shots inside the first 30 seconds, and an average shot length of 2 seconds or under. There is 1 more effect worth naming. Past about 30 seconds, a viewer who has already invested starts to defend that investment, which is why turning a piece around after the 30 second mark is easier than turning it around at 8.

A view is not counted. It is weighed, and it is weighed at 4 checkpoints.

The end checkpoint sorts into 3 outcomes that the feed treats differently: finished, rewatched, shared. Completion is the gate, rewatch is the amplifier, sharing is the multiplier. Completion rate is what short form rewards, which is why a 7 to 15 second piece watched twice outperforms a 60 second piece watched most of the way through 1 time. The shorter piece wins on the arithmetic, not on the idea. 2 rules protect the end. The payoff is never emptied early - the high notes are trailed, never spent at the 4 second mark. And the close carries a call to action rather than trailing off, because an edit that ends without asking for anything ends as 1 view. The probable outcome of a piece that holds 3 seconds with a clear premise and finishes with the promise paid is not a coin flip. The probability that it is promoted rather than buried is high, and it is high for a reason that is visible in the 4 readings before anyone looks at the analytics dashboard. The end is also the cheapest checkpoint to test backwards. Read the last 2 seconds first, then the middle, then the hook. Most pieces fail in the order they were built, which is the reverse.

Format is settled and does not need re-deciding. 9:16 vertical is the default and covers more than 60 per cent of social video. Longer narrative work holds to 30 to 60 seconds only if retention survives every second of it. Trends run on 2 clocks. An audio trend peaks in 7 to 10 days. A format trend - split screen, text overlay, carousel - runs 3 to 4 weeks. The entry window is the first 3 to 5 days, and the signal that a trend is fading is 20 accounts running the same format and brands arriving in numbers. Underneath all of it sits 1 table. Cuts every 0.5 to 2 seconds, most at 1. Captions burned in at 36 to 60 points, keyword highlighted, centred or upper third. Cuts landing on the beat, with 110 to 120 beats per minute mapping to roughly 0.5 second cuts. 2 grammar rules from longer-form work still hold at this length. A cut between 2 shots of the same subject moves at least 30 degrees and changes size, or it reads as a jump cut. And the camera stays on 1 side of the axis, because crossing it flips the direction the viewer believes they are travelling. The one-rule summary runs: hook in the first second, promise by the third, hold every second after with cuts that land on the beat and captions that punch. The same arithmetic has already been proven outside media - 12 countries, 210 energy systems across Africa and Asia and 200 wood-gasification machines across the UK and Europe - where the specification came first and the build was the commodity. The 4 checkpoints are that specification for a 7 second film.