How to write a YouTube hook: the first 15 seconds

The hook is not the first 15 seconds of your video. The hook is the answer to one question: why does this matter right now, and what specific surprise is the viewer about to learn? Every good opening on YouTube is doing those two things. Every opening that fails is doing something else.

Most operators write hooks that describe the topic instead of stating the stakes. "Today we're looking at the collapse of [company]." That is a description. The viewer already knew the topic from the title and thumbnail. The first 15 seconds are not where you restate the packaging. They are where you deliver the first specific surprise that makes the viewer feel the click was worth it.

What a hook must do

A hook has two jobs. The first is to state the stakes: what is the real-world consequence of what is about to be explained. Not the abstract version, the specific version. Not "this company failed," but "eight thousand employees lost their retirement accounts over a decision made in a single meeting." The number and the meeting make it concrete. Concrete stakes hold attention. Abstract stakes do not.

The second job is to deliver a surprise: something the viewer did not already know or did not expect, stated in the first sentence. The surprise is what earns the next 10 seconds. Without a surprise in the opening, the viewer has no reason to keep watching because nothing has differentiated this video from everything else they saw today.

Hooks fail when they do neither of these things. A hook that describes the topic without stakes or surprise is a hook that tells the viewer nothing new. Retention data on channels we operate shows those openings generate a secondary drop between seconds 10 and 20 that compounds the initial scroll-off. The viewers who made it through the first 5 seconds are still deciding whether to commit. The hook either gives them a reason or it doesn't.

Four hook patterns that hold

These four patterns consistently hold retention past the 15-second mark across different niche types and video lengths. Each one works because it delivers a concrete surprise or states stakes in a specific way.

Data-shock

Open with a number or statistic that is hard to walk away from. Not a familiar statistic, a specific one that signals real research.

Example: "In the eighteen months before the bankruptcy filing, the company's executives collectively withdrew $340 million in bonuses. The factory workers were told there was no pension money left."

The data-shock pattern works because the brain registers specific numbers as evidence. The viewer's response is not "interesting" -- it is "wait, what." That response holds them through the next 30 seconds while they wait to understand the number.

It fails when the number is vague or already widely known. "Millions of people use this app" is not a data-shock. "One app processed more financial transactions on a single Tuesday in October 2022 than the US Treasury did in all of 2019" is.

Hypothetical with concrete stakes

Set up a what-if scenario, but anchor it to real-world cause and effect rather than abstract speculation.

Example: "If one pharmaceutical distributor had flagged a single suspicious order in 2012, the opioid crisis in three states looks completely different. They didn't flag it. The order went through. The reason is a legal loophole that still exists today."

The hypothetical creates a knowledge gap immediately. The second sentence collapses the hypothetical into reality. The third sentence promises the gap will be filled. By the time you hit 10 seconds, the viewer has three reasons to keep watching: what was the order, why the loophole, and whether it has been fixed.

The pattern fails when the hypothetical is too speculative to be credible. Scenarios that start with "imagine if" and end with a vague consequence lose viewers because there is nothing concrete to wait for.

Cold artifact

Open on something physical and visible. A document, a receipt, a memo, a screenshot, a photograph. The narration walks the viewer through what it means.

Example: "This is a memo dated March 4, 1999. It was sent internally at a company that no longer exists. The memo describes, in four bullet points, exactly how the company would run out of money within three years. Nobody acted on it."

The artifact gives the viewer a visual anchor in the first frame that is consistent with the title promise. The narration converts that anchor into a story. The pacing is controlled because each sentence reveals one more piece of the artifact.

This pattern is particularly strong for history, investigation, and collapse formats where a physical document or image exists in the archive. It fails if the artifact is not actually interesting or if the narration front-loads too much generic context before getting to the specific detail.

Contrarian summary

Open with the conclusion of the video, stated in a way that contradicts what a reasonable person would expect.

Example: "The most successful logistics company in the world does not own a single warehouse. It has never owned a warehouse. The reason it outcompetes every company that does is one organizational decision its founder made before the company had revenue."

The contrarian summary tells the viewer the topic and the payoff in the same breath. The list of negations builds suspense. The viewer knows where the video is going and is already skeptical in a productive way, which means they are watching to see if the payoff delivers.

This pattern fails when the payoff is ordinary. The contrarian setup raises the bar. If the "one organizational decision" turns out to be something the viewer has heard before, the viewer feels misled and retention drops sharply at the reveal.

Three patterns that fail

These are the most common hook patterns we see on faceless channels. They consistently underperform in retention data, and the reason is the same in all three cases: they delay the stakes or the surprise instead of leading with them.

Date-led open. "On March 14, 2004..." The date is doing no work. It does not state stakes and it does not deliver a surprise. It signals to the viewer that the next sentence will be generic context, and many viewers bail before that next sentence arrives. Whatever fact you were building up to after the date, start with the fact. Dates can appear in sentence two or three.

Question open. "Have you ever wondered why [thing]?" This phrasing has been overused in clickbait for so long that viewers in 2026 process it as noise. The question form also delays the stakes. Instead of asking whether the viewer has wondered something, state the answer as a declarative and let the viewer register the implicit question themselves. The declarative version of the same content is always stronger.

Meta-statement. "In this video, I'm going to explain..." The viewer clicked the thumbnail and read the title. They already have the gist. The meta-statement uses 8 to 12 seconds of the 15-second window to repeat information the viewer already has. The first 30 seconds are where retention is won or lost, and a meta-statement wastes the most valuable portion of it.

How the hook sets up the re-hook

The hook's second function is structural: it creates a promise that the rest of the video is obligated to fulfill. That promise determines where your re-hook goes and what it says.

At the 90-second mark, retention drops again as committed viewers start to drift. The re-hook is what keeps them. But the re-hook can only reference stakes that the hook established. If the hook was vague or generic, there is nothing specific to re-invoke at 90 seconds, and the re-hook falls flat.

A strong hook makes the re-hook easy to write. You already promised a specific payoff. The re-hook is a brief acknowledgment that the payoff is coming, stated in fresh language, with a forward-look tease. If you find yourself struggling to write the re-hook, check the hook first. The re-hook problem is usually a hook problem.

Write the hook and the re-hook as a pair, not separately. The hook creates the knowledge gap. The re-hook reminds the viewer the gap is about to close. That sequence is what carries retention from second 0 to the midpoint of the video.

The practical check

Before finalizing any hook, run it against two questions:

1. Does the first sentence state something specific that a reasonable viewer did not already know? If it does not, rewrite the first sentence.
2. Does the first 15 seconds contain a concrete stake: a number, a consequence, a person, a date, a document, something real? If it does not, it is a description, not a hook.

If both questions get a yes, the hook is doing its job. If either gets a no, the fix is almost always to move the most specific sentence in the opening to the front and cut everything that preceded it. The best hooks are usually buried two or three sentences into a first draft. Find the sentence where something concrete and surprising happens, and start there.

The rest of the script depends on what the hook promises. Start with a promise that is specific enough to be worth keeping, and the structure of the video follows from it. For the research process that feeds strong hooks across any niche, see the niche research guide.