YouTube Shorts Hooks:
The First Two Seconds Decide Everything
A viewer scrolling through Shorts decides whether to stay or leave before your voice finishes its second word. Not after the intro — before it. That means the opening line is not one part of a news video; it is the whole video. Everything the rest of the script does only matters if the opening line earned the next two seconds. A video that is well-researched, well-edited, and beautifully voiced, but opens with a slow line, is a video nobody watches.
This guide is about that line. The three hook formulas that work for news videos, what a hook is actually scored on, and how to test a hook before you publish — instead of after the retention graph tells you it did not work.
Neither is “In this video we look at…”, “Breaking news today”, or “Here is what we know so far.” None of those lines give a scrolling viewer a reason to stop. They give them a reason to scroll.
Why most news Shorts fail in the first two seconds
Most news creators know the opening matters and still write openings that lose. Not because they are careless, but because the instincts that make a good article or a good long-form video are the wrong instincts for a Short.
A written article can open with context. The reader is on the page; they have already committed. A long-form video can open with a welcome, because the viewer clicked on a 20-minute commitment and is prepared to wait. A Short gets neither. The viewer is mid-swipe. Their thumb is moving. The only thing that stops it is a sentence that makes them need to know what comes next, in the half-second before their thumb finishes the swipe.
Here are the four failures that show up most often:
1. Leading with the outlet
“According to Reuters, the EU has set tariffs on Chinese electric vehicles.” The viewer has heard the outlet name and not yet heard the story. By the time the story arrives, the thumb has moved.
What actually happens: The source that makes the video credible is being spent at the exact moment the viewer has no reason to care about credibility. The right place for the outlet name is once, naturally, when the story is already holding attention.
2. Leading with the setup
“Three things you need to know about the Middle East this week.” A promise of value does not count as value. The viewer has to see the value now, not in five seconds.
What actually happens: The video is announcing a video. The scroll moves before the actual content begins. Shorts that front-load the setup do not fail because of the first 10 seconds; they fail in the first 2.
3. Leading with the summary
“The situation in the Gulf remains tense after weeks of strikes and counter-strikes.” This is true, complete, and tells the viewer nothing they can picture. Summaries are what you do when you have already lost the reader and are trying to catch them up.
What actually happens: The hook is technically accurate and emotionally dead. It reads like the second paragraph of an article, where it belongs.
4. Leading with the context that only matters later
“For the past six months, tensions between Tehran and Washington have been rising.” The viewer has no reason yet to care about the past six months. By the time they would have cared, they are gone.
What actually happens: The background is the right material for the third paragraph of the body. In the hook, it delays the payoff that would have earned the viewer’s attention in the first place.
The common thread: each failure postpones the reason to watch. The good hook delivers that reason in the first sentence. Everything else in the script, including the outlet name, comes after. You can read more about how the rest of the script is built around that principle in the fact-checked scripts guide.
The three hook formulas
There is no single way to open a news Short. But there are three openings that reliably work, and a fourth that almost always fails. The formulas are not templates to fill in; they are shapes that the strongest first sentence of your story takes, once you have decided what the strongest first sentence is.
Formula 1 — Consequence first
The most consequential fact, in the plainest words, first. No setup, no attribution, no context. The viewer hears what happened and why it matters to someone who has never heard of the story, in under about ten words. This works when the story has a clean, finished event: a decision made, a strike confirmed, a price set, a vote cast.
Examples of the shape (from real recurring news stories, not fabricated):
- “The EU just set tariffs on Chinese EVs at up to 45%.”
- “Iran has told the US its warships are not welcome in the Gulf.”
- “Russia hit Ukraine’s power grid overnight, again.”
What makes these work: each one names who did what, in a way that a stranger can picture. Each one has a stake: prices, borders, blackouts. None of them open with a source name or a summary. The outlet, the background, the reaction — all of it can come later, if the viewer is still there.
Formula 2 — Question the facts answer
A sharp question that the video answers, when the outcome is still open. This works when the story is in progress: a decision is still being weighed, talks are underway, an election is coming, an investigation is ongoing. The question has to be specific to the story. “What does the future hold for the Middle East?” is a question nobody needs an answer to. “Why would Iran offer to reopen Hormuz now?” is a question that only makes sense if you know a bit about the story — and that is the point: it makes the viewer feel they are about to learn something specific.
Examples of the shape:
- “Why did Beijing wait until now to announce the investigation?”
- “Why would Iran reopen the Strait of Hormuz when it just threatened to close it?”
- “What does Donald Tusk actually get from this deal?”
What makes these work: each one names a specific subject, not a generic theme. The viewer hears the actor’s name and the tension in the same breath. The question sounds like something a curious person would ask, not a headline writer. And the video’s body answers it with facts, not with more questions.
Formula 3 — The hidden layer
A contrast, or a hidden layer, that reveals the story is not the one the viewer thinks it is. This works when the headline is true but incomplete — when the interesting part of the story is what the coverage is not saying. It is the most dangerous formula to write badly, because it can slide into clickbait if the payoff is not there. It is the most powerful formula when it is written well, because it makes the viewer stop and re-read the situation.
Examples of the shape:
- “Everyone is watching the front line. The real move is somewhere else.”
- “The sanctions story is not about the sanctions.”
- “Poland says it is helping Ukraine. The numbers say something else.”
What makes these work: each one delivers a contrast inside two sentences. The viewer’s first reaction is “wait, what do you mean?”, and the second is “I need to find out”. The payoff has to be in the body — a specific fact, a specific number, a specific quote that makes the contrast real. If the video does not deliver that, the hook was clickbait, and the viewer’s next reaction is to leave and not come back.
A story where the outcome is decided and specific is a consequence-first hook. A story where the outcome is still open is a question. A story where the headline hides the point is a hidden layer. When in doubt, ask: which of those three sentences can I write in under 13 words, without inventing anything, that makes a stranger who has never heard of this story need to keep watching?
How Autonom scores a hook
A hook is not a matter of taste. Whatever a creator thinks about their opening line, the audience reacts on the same terms — did it stop the scroll, did it make them need the next sentence, did it promise something the video pays off? You can score those properties mechanically, before you publish, and use the score as a check on your instincts.
Autonom Newsreel uses a small rubric, roughly the one below. Every hook written is scored, and the score is shown alongside the hook in the review step. The score is not a verdict; it is a test that catches a hook that is clearly weak, so you can rewrite it before you spend a video on it.
Short
A hook under about 9 words is short; up to 13 is tight. Over 16 words and the viewer has had to think about the sentence instead of about the story.
Why it matters: Every word in the hook is a word the viewer has to spend attention on before they decide to stay. Fewer words, more time to feel the stake.
Specific
A hook with a number, or a capitalised name that is not the first word, reads as specific. “Up to 45%.” “Donald Tusk.” “The Strait of Hormuz.” Specific words are what make a viewer feel the story is real.
Why it matters: The alternative — “the situation in the Gulf”, “the tariffs on cars”, “officials said” — is the language of a summary, not a story.
Real stakes
A hook that mentions prices, borders, blackouts, war, sanctions, deaths, elections, or a specific commodity is talking about something the viewer can feel. These are the words that make a scrolling viewer stop.
Why it matters: A story about procedure — a hearing, a commission, a probe — is not a story that stops the scroll. A story about a consequence is.
Curiosity gap
Words like secret, hidden, quietly, no one, nobody, without warning, just, suddenly, but, despite, only, behind signal that there is something to find out. They open a gap the viewer has to close.
Why it matters: The gap has to be real. A hook that uses curiosity words to promise something the video does not deliver is a hook that loses the audience in the next 30 seconds, not the first 2.
Opens with who, not the outlet
A hook that starts with an outlet name, “according to”, “per”, “in this”, or “breaking” is scored down. The source is a reason to trust the video, not a reason to watch it.
Why it matters: The outlet name is a signal to a viewer who is already convinced. For a viewer mid-swipe, it is dead air.
No weak fillers
Words like basically, essentially, actually, very, really signal a hook that has not decided what it means. They are not fatal; they are a warning sign that the sentence can be sharper.
Why it matters: A hook that hedges its own importance is a hook that has already told the viewer it is not important.
None of those categories is a creative choice. They are properties of the sentence, and you can measure them. The rubric is intentionally narrow so that ordinary variation survives it: not every good hook hits every category, and not every hook that hits every category is good. But a hook that scores badly across the whole rubric is almost never one that holds a viewer past the first two seconds, and that is what the score is for.
How to test a hook before you publish
You have written one hook. The temptation is to publish it as is, then read the retention graph and decide whether it worked. That is the wrong way around. A retention graph tells you what happened after the viewer left; it does not tell you which alternative hook would have kept them.
The right way is to write at least three candidate hooks before you decide, and to test each one against the same story before you pick. That is what Autonom does, and it is what you can do manually in a text file if you are writing yourself.
- Write one hook in each of the three formulas. Not three consequences. One consequence, one question, one hidden-layer. Then read them next to each other. Very often the one you instinctively wrote first is not the strongest of the three; the formula forced you to see what you were missing.
- Score them on the same rubric. Short, specific, real stakes, a curiosity gap, who not the outlet, no weak fillers. The score does not pick the winner; it rules out the clearly weak.
- Fact-check each candidate. A hook is a sentence the audience will hear first. If it contains a number that is not in the source, a name that is not in the source, or a claim the source does not make, it is not a hook — it is a fact error on the front page. Every candidate has to pass the same checks as the rest of the script before it can be used. The fact-checked scripts guide covers what those checks are.
- Swap and re-read. Once you have two or three candidates that pass, swap them in and read the whole script with each one. The hook is not a standalone sentence; it is the first sentence of a story. A hook that sounds great alone but throws off the pacing of the body is the wrong hook.
In Autonom, this is a button. Every script comes with two alternative hooks, each in a different formula from the one the model chose, each scored and each fact-checked against the article before it is offered. If you click one, the tool re-audits it — the same sentence-level audit the rest of the script gets, so you cannot accidentally swap in a hook with an unsupported number or a wrong name. If the audit finds a problem, the swap is rejected and the old hook stays. You can see the swap in the review step, alongside the hook score and the suggested alternatives.
The same principle extends past YouTube. Instagram Reels, TikTok, X, LinkedIn — the first line of a caption plays the same role as the first sentence of the video. A caption that opens with a summary loses the viewer before they read the body. A caption that opens with the story’s strongest consequence, or its sharpest question, is the one that gets read. The publishing guide covers how to write the caption per platform, but the shape of the opening line is the same everywhere.
Here is what a bad hook looks like next to the same story with three scored alternatives. The scores are on the same rubric described above; the fact-check is the same one every script gets.
Score 26 / 100. Opens with the outlet (−18), 25 words (too long, −12), no number, no name, no stakes, no curiosity gap. The viewer has learned nothing they can picture.
Question: “Why would Iran offer to reopen Hormuz when it just threatened to close it?” (Score 82. Sharp question, curiosity gap, specific name, no outlet.)
Hidden layer: “Everyone is watching the warships. Iran’s real message is for Beijing.” (Score 79. Contrast inside two sentences, specific name in the second, curiosity gap. Only works if the body actually explains the message to Beijing — which the facts have to support.)
The bad hook is true, complete, and tells the viewer nothing. Each of the scored hooks is a sentence a scrolling viewer has to think about for a second — and that second is the whole video.
Three hooks per script. Scored and fact-checked before you pick.
Autonom Newsreel writes the opening line in the formula the story’s shape calls for, offers two more in the other formulas, and fact-checks each candidate before it is offered. Free to run on your own API keys.
Open Autonom Newsreel →Related guides
Frequently asked questions
How long should a YouTube Shorts hook be?
Under 13 words is the target. Under 9 is stronger if the story allows it, because every word is attention the viewer has to spend before they decide to stay. A hook that runs over 16 words is almost always a summary that has been mislabelled as an opening. The number is not a hard limit; it is a proxy for “does the sentence deliver the reason to watch before the viewer’s thumb finishes the swipe?”
What is the best hook formula for news Shorts?
There is no single best formula, because the formula follows the story. A finished event — a decision made, a strike confirmed, a tariff set — is a consequence-first hook. An open outcome — talks underway, an election coming, an investigation ongoing — is a question. A headline that hides the point — the coverage focused on the wrong thing, the interesting part barely mentioned — is a hidden layer. Write one candidate in each of the three formulas before you pick. Very often the strongest of the three is not the one you instinctively wrote first.
Is “According to Reuters” ever a good opening?
Almost never as the first sentence. The outlet name is a reason to trust the video, not a reason to watch it. A viewer mid-swipe has not yet decided whether to care about the story, so they are not yet in the market for a source. Put the outlet name once, naturally, at the point in the body where the story has already started to hold attention. “CBS reports” as the second sentence, after the event has been stated, is fine. “CBS reports” as the first sentence, before the event, wastes the two seconds that decide whether anyone hears the rest.
Can I test a hook without publishing?
Yes, by writing at least three candidates and swapping them into the same script. Read the whole script with each one — not just the opening line, because the hook is the first sentence of a story, not a standalone. The best hook alone can still be the wrong hook if it throws off the pacing of the body or the flow into the closing question. Autonom does the same swap in the review step, with a fact-check on the new hook before it is accepted.
What does a hook score of 80 out of 100 mean?
It means the hook hits most of the rubric: short, specific, real stakes, a curiosity gap, opens with who and not the outlet, no weak fillers. It does not mean the hook is good — that is still a judgement about the story. A hook that scores 80 can be boring; a hook that scores 60 can be great because it does something the rubric does not name. The score is a floor, not a ceiling: it rules out the clearly weak hooks and lets you focus your judgement on the ones that could genuinely work.
Do hooks work the same way on TikTok and Instagram Reels?
The core principle is the same: the first two seconds decide whether the viewer stays. The shape of the hook changes slightly because the audience behaviour does. TikTok viewers tend to swipe faster than YouTube Shorts viewers and to reward surprise more; Instagram Reels viewers respond more to a first line that feels relatable. The three formulas still apply, but the same story might lean question-first on TikTok and consequence-first on YouTube. The publishing guide covers how to write per platform.