Key takeaways
- YouTube Studio's own audience-retention report labels the first thirty seconds of every video as the Intro, and puts your video into the "above typical" bucket when at least half of the audience is still watching at the thirty-second mark. That is the only public retention threshold YouTube commits to on paper.
- From 24 August 2026 a view is counted the moment playback begins. The click no longer proves anything about the video — the number that carried meaning has moved to engaged views, and the boundary between an impression and an engaged view sits inside your hook.
- Native A/B testing, YouTube's Test & Compare, picks the winning thumbnail on watch time rather than click-through rate. A packaging change that lifts CTR but breaks the intro loses the test it was invented to win.
- A hook is not an opening, an intro card, a logo animation or "hey guys welcome back". Its three jobs are to pay off the thumbnail, prove there is more and install one specific question the rest of the video is going to answer.
- The one line from MrBeast's leaked production handbook that survives outside his budget is that the first minute is the most important minute because that is where the audience decides. It applies to a channel of any size, and it is measurable inside the free analytics you already have.
Almost every post on this blog so far has been about the click. Thumbnails, titles, feed grids, colour, contrast, faces — all machinery for winning a rectangle of screen the eye passes over in a second. What is under-discussed, and what decides whether any of that work compounds, is the thirty seconds that begin the moment the click lands.
That window is the least glamorous part of a video and the highest-leverage part of a channel. It has its own metric in YouTube Studio, and, since August 2026, its own new job — a "view" now happens on the first frame, and the only thing separating a view from an engaged view is what the viewer decides in the opening window. Everything downstream — session watch time, satisfaction, distribution — is priced off that decision.
What YouTube itself calls the "Intro"
YouTube's audience-retention report is not a general graph — it is a labelled document. Inside YouTube Studio, on any video, the report highlights four types of moment on top of the retention curve: an Intro, one or more Top Moments, Spikes where sections were rewatched or shared, and Dips where a lot of viewers left in a short window. Each has a definition in YouTube's own Help documentation and each is worth reading against.
The Intro is defined as the percentage of your audience still watching after the first thirty seconds — not the first minute, not the first ten. Thirty seconds is the reference window YouTube's own analytics use, and it is the one interval you can measure honestly across every video without normalising for length. The Help page also gives the only public retention threshold YouTube commits to: a video whose Intro figure sits at fifty per cent or above is placed in the "above typical intros" bucket. Every other benchmark you read online — "aim for 70% at 30 seconds", "35–75% average retention is good" — was invented by somebody. The 50% mark is the one that came out of YouTube.
Two things follow. First, if you are going to sort your own videos by a single retention number, the thirty-second Intro figure is the one YouTube itself categorises your work by. Second, the report is telling you where to look. A steep drop inside the first thirty seconds is not one problem out of many. It is the problem, filed under its own heading, on the page YouTube built to help you diagnose the video.
Where to find the Intro number
In YouTube Studio, click Content, click into a video, click Analytics on the left rail, then the Engagement tab. Scroll past the average-view-duration card to Audience retention. The four-moment labels sit on top of the curve; the Intro card sits above it. On new videos it can take one to two days to appear.
Why the intro matters more since 24 August 2026, not less
Until August 2026, a long-form view was only counted after roughly thirty seconds of watching — the public view count had a de-facto retention filter baked into it. From 24 August 2026 a view is counted from the first frame of playback, the same way Shorts have been counted since March 2025. The two-number system that replaces it is a looser public view count and an engaged view — a play where the viewer stayed past the opening seconds or actively did something (a like, a comment, a share) — which is what YouTube now treats as the "real" number and what still governs monetisation.
The old thirty-second threshold used to happen in silence: the click became a view or it did not, and the boundary was hidden inside the counter. Now the boundary is exposed. Two videos with identical impressions and identical click-through can produce very different revenue and very different distribution, and the split happens inside the opening seconds. The public number tells you how many people saw the thumbnail hard enough to press play. The engaged view tells you how many stayed long enough that the click was worth having.
Creators who used to talk about their opening as a design choice now talk about it as an economic one. If your intro consistently converts fewer than half of the plays into engaged views, the channel's growth curve gets flatter irrespective of how good the thumbnail work is. For the mechanics of the wider view-counting shift, see our post on what a good CTR actually is on YouTube.
The three jobs of a hook
Everything a hook has to do fits into three responsibilities. If any of them is missing, the Intro number underperforms, and no amount of pacing or B-roll rescues it.
1. Pay off the thumbnail — fast, and visibly
The click was won by a specific promise. Show it being kept early enough that the viewer's expectation and the video's opening frame line up. If the thumbnail showed a demolished car, the demolished car appears in the first fifteen seconds; if the thumbnail showed a number, the number is spoken before an intro card would have finished playing. "Delivering on the thumbnail" is the first failure mode YouTube's own Help page names when it explains why an intro loses viewers — a diagnosable defect, not a stylistic complaint.
2. Prove there is more
The second job is to establish that the thumbnail's promise is not the whole video. Every strong hook makes this argument silently — "there is a bigger version of the thing you clicked for, further in." A cold open that hands you the whole answer in fifteen seconds wins nothing, because the viewer's next act is to close the tab satisfied. The intro has to leave a specific gap it can only fill by playing on.
3. Install one specific question
This is where most hooks fail. A good intro plants exactly one unanswered question the rest of the video will answer, and does it explicitly enough that the viewer can name what they are still watching for. Vague promises — "wait until you see what happens next" — signal a question rather than installing one. George Loewenstein's information-gap work is consistent on this: the pull of an unanswered question is proportional to how specific the missing piece is. "Which of these three did the trick" is stronger than "you'll see". Our post on the psychology of the click goes further into why.
What a hook is not
The other half of the definition is worth stating flat. A hook is not:
- An intro card, a channel-branded sting or an animated logo. Pure negative retention — they buy nothing and cost every viewer who came in cold. A channel with a fixed intro card loses roughly the same slice of audience in the same window on every upload.
- A greeting — "hey guys, welcome back to the channel". The video equivalent of clearing your throat. In a feed where the viewer already committed a fraction of a second to give you attention, a warm-up burns it.
- A self-introduction. If it's a subscriber they know; if it's a first-time viewer they will find out from the second sentence onwards. The self-intro is a legacy of television, where the host had to identify themselves because the channel identifier was small. On YouTube the channel identifier sits under the video.
- A preview reel. Contested; works for some genres, fails in most because it either spoils the payoff or gives the viewer permission to sample two seconds of each highlight and leave.
- The sponsorship read. A pre-roll sponsor read is a retention grenade in the exact window YouTube measures for the Intro metric. Sponsors belong at the first natural pause, not before the hook has had a chance to work.
The anatomy of the first thirty seconds
Splitting the window into three sub-intervals makes the design problem easier, because each has a different job.
0–3 seconds — the visual commitment. The screen changes from whatever the thumbnail was to something that continues that promise. If the thumbnail was a face reacting to something, the video opens on that face; if the thumbnail was a number, the first shot contains it. A mismatch here is felt as a micro-betrayal, and the retention curve almost always shows a small cliff at second two or three — well before the intro card the creator suspects.
3–10 seconds — the payoff sentence. One sentence, spoken plainly, that establishes what the video is going to give the viewer. "The reason it fell is not what the title said." "Here are the seven mistakes I made rebuilding this, in order of expense." Every working example contains a specific object or number — not "we'll be looking at some interesting things".
10–30 seconds — the commitment. The last twenty seconds is where you install the specific question and, if the format allows, prove there is stakes or scale beyond the opening frame. Not "in this video we will discuss", but "before we start, the third one is the one that actually worked, and none of the first two look anything like it". By second thirty the viewer should be able to say what they are still watching for. If they cannot, they leave. That is the audible content of the Intro metric.
The hook shapes that actually work
The catalogue of "hook types" published online is longer than it needs to be. Underneath the names, there are roughly six shapes a working hook takes on YouTube, and they map cleanly onto video categories.
The cold open
Drop the viewer into the middle of the story, with no set-up. The first shot is not a title card — it is the moment before the outcome. This is the shape MrBeast's leaked production handbook, published in September 2024 and later confirmed by former producers, spends most of its retention discussion on: pay the video off in the first minute because the first minute is where viewers decide. It works best on stakes-driven video and documentary formats. It works badly on tutorials, where the reason for watching is the sequence itself.
The proof-first hook
Open on the outcome, then work backwards. If the video is about a transformation, the finished version is on screen inside five seconds; if the video is about a claim, the evidence appears first. This is the shape most before-and-after content should use, and most does not — creators save the reveal for the end, which flatters their editing instinct but starves the retention curve of a reason to keep watching.
The contrarian claim
A single, specific claim that runs against consensus, made in the opening sentence and then defended. "The most popular piece of advice on this topic is wrong, and here is what the numbers actually show." Works on analysis, review and educational content. Fails on anything the audience does not already have a prior belief about, because there is nothing to contradict.
The stakes hook
Establish immediately what is being risked. "I have twenty-four hours to do this and my livelihood depends on the result." Stakes are the simplest way to create the sense of "more to come" that the second job of a hook requires. The common failure is making them abstract — "big consequences" is not a stake; "if I do not make this shot my brother eats the chilli" is.
The question hook
Ask a specific, un-obvious question in the opening sentence and refuse to answer it until later. "Why do the best-performing videos on this channel all have the same one thing in common?" beats "have you ever wondered why some videos do better than others?" by a large margin, because the first admits a real, checkable answer exists.
The tutorial-start hook
On instructional content, the working hook is almost the opposite of the cold open: skip the ceremony and start the demonstration. First shot, the first step, on screen. The premise, the software and the reason fit inside a two-sentence voiceover that plays over the first action. Every second of "before we begin, let me show you what we're going to build" erodes the tutorial's Intro number.
How to read your retention curve
The audience-retention graph is the most honest document YouTube gives you about your work, and it is under-read because most creators only glance at the average. The four labelled moments YouTube overlays on top of the curve — Intro, Top Moments, Spikes, Dips — are a diagnostic key.
The opening cliff. A sharp, near-vertical drop inside the first ten seconds almost always means one of two things: the thumbnail promised something the opening did not deliver, or the video begins with an intro card, sponsor read or self-introduction that gives unengaged viewers permission to leave. The fix is editorial. Cut everything before the first line that pays off the thumbnail.
The slow-slope intro. If the curve does not cliff but slopes downward across the first minute, the hook is not failing in a specific place — the video has simply not established its promise. This is the "and today we're going to be looking at" shape, and the fix is rewriting the opening voiceover, not re-editing the picture.
Recovery bumps. A dip that then recovers means viewers are skipping ahead, usually because the section felt slow. Cut it, or compress it, and the average lifts.
Top Moments and Spikes. A Top Moment is a section almost nobody left during; a Spike is a moment that was rewatched or shared. Spikes are the closest proxy YouTube has to satisfaction without asking, and they tend to cluster on the payoff moment. If the current video has a spike but the previous one did not, the difference is often whether the promise was clearly enough framed that the viewer knew when it had been delivered.
The one comparison worth running
Take your five best-performing videos and your five worst-performing videos and open the audience-retention report on each. Ignore the average. Look only at the shape of the first thirty seconds. In nine cases out of ten the difference is not in the middle of the video — it is in the opening curve. The good ones cliff less; the bad ones cliff early and the whole curve is displaced downward for the rest of the video.
Why watch time beats clicks now, and what that does to hook design
The last five years of YouTube advice were mostly written under an assumption that no longer holds — that click-through rate is the single most important number to maximise. Two shifts have quietly rewritten that.
The first is that Test & Compare, YouTube's own thumbnail and title A/B testing tool, picks the winning variant on watch time per impression rather than click-through rate. Between three variants of a thumbnail the one selected as the default is the one whose viewers stayed longest, not the one that got the most clicks. That is a public change in policy — YouTube's own Help page says the tool "optimises for overall watch time over other metrics like CTR" — and it means a thumbnail change that lifts clicks but degrades the intro is being marked as a loser by YouTube's own machinery. Our deep dive on Test & Compare covers the mechanics.
The second shift comes from repeated public comments by YouTube's own team. Todd Beaupré, Senior Director of Growth and Discovery, has said in Creator Insider and Creator Liaison appearances that the recommendation system is trying to understand not just what a viewer watches but "how they feel about the time they're spending" — and that creators should "replace the word algorithm in your question with audience". Satisfaction is measured through post-watch surveys, likes, dislikes and rewatch behaviour, and it is being weighted more heavily each year. On the hook side that translates simply: an opening that oversells and under-delivers hurts you twice — once through the retention drop, again through the satisfaction score that hangs on the survey and rewatch behaviour of the viewers who stayed.
The two intro failures nobody talks about
Most published advice on hooks concentrates on the case where the intro is too slow. There are two other failures worth naming because they are quieter and, in each case, look virtuous on the surface.
Payoff before commitment
The over-corrected hook delivers the entire answer in the first thirty seconds and then spends the rest of the video backfilling. This produces a great Intro number and a terrible average view duration — the viewer got what they came for and left. The fix is not to slow the hook down; it is to widen the aperture. The intro gives the shape of the answer without the detail, so the viewer who came for the answer stays for the specifics. If the intro is a summary of the ending, the ending is redundant.
Mismatch to the thumbnail, delivered too late
A subtler failure than the intro that never matches. The intro does match, but it takes forty seconds to get there — and by then the Intro window has closed and the audience has dropped by a third. This is common on well-produced videos where the creator prefers a stylish opening: an atmospheric shot, a bar of music, a slow zoom into the environment before the real payoff. If you cannot cut it for taste reasons, move the audio content of the payoff over it. The picture can breathe; the voice cannot afford to.
A practical framework for writing a hook
Everything above is diagnostic. Here is how to apply it to the next video you upload.
- Write the thumbnail promise as one sentence, in full. The exact claim the thumbnail is making — object, outcome, number. This is your hook contract.
- Write the payoff sentence. Forty words or fewer, delivering the shape of the thumbnail promise. Recognisable to a viewer who just saw the thumbnail — same object, same specific.
- Pick one of the six hook shapes. Cold open, proof-first, contrarian, stakes, question, tutorial-start. Shape follows the format, not personal preference.
- Write the specific question the video will answer. If you cannot state it in a sentence, you do not have a question — you have a topic, and the intro will feel vague.
- Script the first thirty seconds and time them. A hundred and twenty spoken words is roughly thirty seconds. If your opening scripts run to two hundred, the last hundred are eating your Intro number.
- Record two intros, cold. The scripted one, and a second version starting thirty seconds later — as if the first thirty seconds had been cut. Try both in the edit.
- Check the Intro number at forty-eight hours. If it is under fifty per cent — the "above typical" line YouTube uses — the next video's intro gets rewritten based on where in the opening curve the drop happens. This is the highest-payoff feedback loop on the channel.
Testing an intro without reshooting
You cannot Test & Compare an intro directly — YouTube's A/B tool covers thumbnails and, since 2025, titles, not video content. There are two workable substitutes.
The first: run the thumbnail test knowing what you are looking for. Because Test & Compare picks the winner on watch time, a thumbnail variant that changes what the video seems to promise indirectly measures whether the hook can carry that promise. A bolder thumbnail that lifts clicks but shows up as "Performed Same" or "Inconclusive" usually means the intro cannot deliver on what the bolder thumbnail promised. Our complete thumbnail A/B testing guide covers the reading of those outcomes.
The second: swap thumbnails and titles on old videos where the packaging was the problem, not the intro. Fixing the packaging exposes the intro to fresh traffic, and if the newly-arrived viewers cliff early, the intro needs work too. Our post on changing a thumbnail after upload covers the mechanics of that swap.
The Shorts case is different
Everything above applies to long-form video. Shorts are their own case: playback starts automatically, the swipe-away happens in the first second or two, and the retention window that matters is closer to three seconds than thirty. On a Short the equivalent of the hook is a single visual moment and a single line of on-screen text, and there is still no A/B tool for it. Custom Shorts thumbnails, rolled out in July 2026, changed the packaging half but not the retention half. Our Shorts thumbnail guide unpacks the difference; the short version is that pacing, mismatch and stakes apply, compressed into a tenth of the time.
What all of this quietly makes true
Hook discipline compounds in a way thumbnail wins do not. A thumbnail change wins the impression once, and if the video does not hold the audience the algorithm rebases on the retention it sees. An intro that reliably converts plays into engaged views trains the system's model of the channel: it starts recommending your videos to viewers more likely to stay, and the whole flywheel starts to run in your favour instead of against you.
The uncomfortable part is that hook discipline is boring work. Thirty seconds of video, written and rewritten and timed, on a video that already felt finished. Few creators run a habit of watching their own opening back with the sound off and asking whether the picture confirms the thumbnail. Fewer still cut a stylish opening they are proud of because the Intro number said the audience did not stay for it. That is the split between channels that grow through 2027 and channels that keep wondering why the impressions keep coming but the views keep softening.
If you want a place to start, open your five most recent videos, look at the Intro figure on each, and note which of the six hook shapes the opening actually used. The pattern that emerges is usually not the one you would have guessed. The videos that won held the audience because the first thirty seconds paid off the thumbnail without asking the viewer to earn it, and installed the one specific question the rest of the video was there to answer.
The rest of the machinery — packaging, colour, format, session strategy — sits on top of that intro. Our complete thumbnail playbook covers the packaging side; the views hacks post works down through every lever a channel has. And when you are ready to package a thumbnail that actually delivers on the promise your intro will pay off, Thumblore is built for the version of the job you cannot afford to do slowly.