Everyone teaches the hook. Almost nobody teaches what comes next.
There are five hook patterns worth knowing, and you can find templates for all
of them in ten seconds. What you cannot easily find is the part that decides whether the hook
worked: the fifteen seconds immediately after it. That is where a good opening
quietly loses the viewer, and that is what this book is about.
Chapter 1
You already paid for the click
The most expensive thing your video ever does has happened before a single frame plays.
Someone was scrolling. Thousands of things wanted their attention and were losing. Your thumbnail and your title took that person, stopped them, and made them tap. Whatever you think of your packaging, that was the expensive part: you spent an idea, a title you rewrote nine times, and a thumbnail you are still not sure about, and it worked.
And then a great many videos spend the next fifteen seconds saying hello.
The click is bought. The opening decides whether you keep what you bought.
It helps to see these as two separate economies. The thumbnail and title buy attention. The opening decides whether you keep it. They are bought and spent in different currencies, which is why a creator can be good at one and quietly terrible at the other for years without noticing, because the click-through rate looks fine and the videos still do not grow.
The second economy is the harsher one. A viewer who never clicks costs you nothing. A viewer who clicks and leaves at eleven seconds costs you something real: you now have a video that people opened and abandoned, and that is a worse signal than one nobody opened. Getting better at thumbnails while the opening stays weak does not fix the channel, it scales the leak.
Nothing in this book is about getting more clicks. It assumes you already earned them, the hard way, and it is about the fifteen seconds where most of them are given back.
Chapter 2
A hook is not an intro
One of these is a sentence about your video. The other is a sentence about the world.
Ask one question about your opening line: am I hooking, or am I promising?
Promising is announcing what you are about to do. "Today I am going to show you three ways to fix your audio." Nothing in that sentence is unresolved, so nothing pulls. It is a sentence about the video.
Nobody arrived wanting to hear about the video. They arrived wanting the thing the video is about.
Hooking is delivering something that opens a gap. "My audio sounded terrible for two years and one setting was doing it." Same subject, same information, and now there is a specific missing fact. It is a sentence about the world.
Be careful with the word promising here, because it is doing a narrow job. It does not mean the promise your title and thumbnail made. That promise is a debt you absolutely must pay, and paying it visibly in the first fifteen seconds is one of the things a good opening does. What is being warned against is the announcement: the sentence that says I am going to tell you instead of telling you.
The tell is grammatical, which makes it easy to catch in your own script. Look at the subject of your first sentence. If the subject is this video, this channel, or today, you are promising. If the subject is a thing that happened, a number, a person, or a claim, you are hooking. Every chapter in Part II ends with four rewrites in exactly this shape, because seeing your own sentence in the left column is faster than any rule.
There is one honest exception, and it matters. Somebody who searched "how to tie a bowline" wants the knot, not suspense. On a search-intent video, announcing the agenda and then delivering it immediately is correct, and manufacturing a gap between the viewer and the thing they came for is just friction. Hook when they are browsing. Answer when they are asking.
Chapter 3
The follow-up is the hook
The opening line is not the hook. It is the first beat of one, and it is not the beat that decides.
Almost everything written about hooks is written about the first sentence. That is the part that is easy to collect, easy to list, and easy to sell as fifty swipeable formulas. It is also the part that has the least to do with whether the video holds.
A hook is better understood as a debt. The opening takes something from the viewer on credit: their attention, on the promise that it will be worth it. What you say next is the repayment. A strong opening followed by a wandering forty seconds does not fail gently, it fails worse than a flat opening, because you raised the price and then did not pay.
A hook is a debt. The only interesting question is how you pay it.
This is visible in your own analytics, and it is the single most useful thing to go and look at after reading this. Open a video and look at the first minute of the retention graph. A drop in the first three seconds is a packaging problem: the thumbnail promised something the video is not. A strong first three seconds followed by a cliff somewhere between fifteen and forty-five seconds is almost never a hook problem. It is an unpaid hook.
The five patterns in Part II are not five ways to write a sentence. They are five different debts, and the reason each one gets its own chapter is that they are not paid in the same currency. A result hook is paid with proof, because the viewer's real question is whether what you just showed them is staged. A stakes hook is paid with relief, because a named risk with no way out is a reason to leave rather than to stay. A contrarian hook is paid with evidence, and under more pressure than the rest, because the viewer is not merely waiting, they are already assembling a rebuttal. A pattern break is paid with orientation, because confusion cannot become interest until the viewer knows what they are looking at. An open loop is paid partially and quickly, one real piece of the answer handed over so the viewer learns that you actually have the rest.
Get the currency wrong and it does not matter how good the opening line was. That is the whole book, and everything after this is detail.
Chapter 4
Choosing a pattern
Not from a list of five. From what you actually have.
Nobody picks a hook by reading five options and feeling which is nicest. You pick it from two things you already know before you write a word: who this video is for, and what you have actually got.
The first question is whether the viewer searched or was scrolling. Someone who typed a question is owed an answer, quickly, and most of the interesting openings are wrong for them. Someone who was scrolling owes you nothing at all and has to be stopped. The same video idea takes a different opening depending on which of those you are expecting, which is why this is the horizontal axis below.
A pattern that fits everywhere is the one most likely to be reached for out of habit.
The second question is what the hook can run on. Some videos contain a thing you can put on screen and let the viewer judge: a finished object, a number, a receipt. Others contain something that has to be felt before it means anything: a risk, a surprise, a reversal. Evidence and emotion are not a quality ranking, they are two different materials, and a hook built from the one you do not have is the most common way a good idea opens badly.
Two cautions about reading the chart. The open loop sits in the middle because it genuinely fits almost anywhere, and that is as much a warning as a permission: the pattern that fits everywhere is the one most likely to be reached for out of habit rather than judgement. And nothing here is a rule about quality. A pattern break on a search-intent tutorial is not a worse hook, it is the wrong material for the job, in the way a wrench is not a worse hammer.
The second figure is the one worth keeping. It is this whole book compressed: the patterns differ less in how they open than in how long you have before the opening has to be paid, and in what it has to be paid with.
Figure 1 Which pattern fits what you have
Result first Shows a thing that happened. Works on people who came to learn how, and it argues with an image rather than a feeling.
The stakes Lives in money, health and safety, which are searched more than browsed, but it runs on exposure rather than on evidence.
Open loop Sits in the middle because it genuinely fits almost anywhere. The exception is the one place it hurts: a search-intent video.
Contrarian Nobody searches for your opinion, so it lives in the feed, but it collapses without proof. Belief is fought with evidence.
Pattern break Pure feed. It works on people who were not looking for anything, and it runs entirely on a reaction to an image.
Figure 2 How long you have before it has to be paid
The bar is the time you have, not the time it takes. Longer is not better:
the open loop and the pattern break have to move fastest, and the stakes hook is the only
one that can hold a viewer for forty-five seconds on tension alone.
Part two · The five patterns
Pattern
Result first
Show the destination first. The gap you open is not what happened, it is how on earth you got there.
How it actually works
A result hook works by removing the outcome from the equation. The viewer already knows how it ends, so the only thing left to want is the road, and that is a much more patient kind of curiosity than suspense.
It only works if the result is legible in a couple of seconds. If the viewer has to be told why the result is impressive, it is not a result hook, it is an explanation with a photo attached.
The counterintuitive part is that spoiling the ending costs you nothing. People rewatch films they know the ending of. What they will not do is wait through a preamble to find out whether the video is worth their time.
Use it for
Builds, transformations and restorations
The after is visible in one frame and the labour is invisible, which is exactly the shape this pattern wants.
Anything with a number attached
A number is legible instantly and carries its own scale. "It sold for 40,000" does the work of a paragraph.
Skill demonstrations
Competence reads immediately on camera, and how someone got competent is a genuinely open question.
Do not use it for
Results your audience cannot evaluate
If they cannot tell good from bad at a glance, the hook lands as a shrug. A niche result needs a comparison in the same breath or a different pattern entirely.
Videos where the result is the only content
If the how is thin, opening on the result finishes the video in eight seconds and the retention graph will show exactly that.
The part that decides it
What you say next
A result hook is paid with PROOF, and this is where most of them fail. You show the after, and the viewer half believes it. Everything you say next either converts that into belief or loses them. What kills it is the instinct to explain yourself first: showing the result and then saying "so, a bit of background on how I got into this" spends the belief you just earned on nothing.
0:00 to 0:05
Show the result. Do not narrate it while you show it.
WhyLet the image land on its own. A voice explaining what you are looking at implies the picture is not carrying it, and the viewer believes the implication.
0:05 to 0:20
Prove it is real. Show it in context, in one continuous shot, being used, or with the thing it is being compared to in frame.
WhyThe unspoken question is not "how" yet, it is "is this staged". Proof answers the question they have not asked out loud, and until it is answered they are not really watching.
0:20 to 0:45
Name the obstacle. Say the one thing that should have made this impossible.
WhyThis converts a result into a story. Without an obstacle, the how is a process video. With one, the viewer now needs to know how you got past a specific problem.
The test: Cover the first five seconds and watch from there. If the rest still makes sense, your hook was decoration rather than a setup, and the video does not actually need it.
Hooking, or promising?
Promising here means announcing what you are about to do. (Different from your title's promise, which you should keep.) With a result hook, promising usually shows up as describing the result instead of showing it.
PromisingI spent three months restoring this and I want to walk you through the process.
HookingThis is the chair. Three months ago it was in a skip.
The left explains the video. The right puts the before and after in one sentence and lets the images do the rest. Notice the right is shorter, which is usually the tell.
PromisingThe results of this experiment were honestly pretty surprising.
HookingNine of the ten failed. The tenth is still running.
Adjectives about a result are not a result. Say the number and the surprise arrives on its own; call it surprising and the viewer decides whether to agree.
PromisingBy the end of this video you will see the finished kitchen.
HookingThat is the finished kitchen. We had eleven days and no plumber.
The left withholds the payoff and asks for patience. The right spends it immediately and buys a better question, which is the whole trade this pattern makes.
PromisingI managed to grow the channel a lot faster than I expected.
HookingForty thousand subscribers in five months, and I posted eleven times.
Vague scale is not scale. Two numbers that sit oddly together open a gap that "a lot faster" cannot.
Running it without your face
This is the strongest pattern for faceless channels, because the result is an object, a screen or a number and none of those need a person to sell them.
The object, shot properly. One clean well-lit frame of the finished thing beats any sentence about it.
A hard cut from before to after, held long enough to be read. The cut IS the argument.
A number on screen, said aloud at the same time. Two channels for one fact is not redundant, it is insurance against a muted phone.
Hands. Faceless does not mean personless, and hands doing skilled work carry credibility a voice cannot.
The risk: The proof beat is harder without a face, because you cannot use your own reaction as evidence. Compensate with continuity: an unbroken shot of the thing working is the strongest proof available to a faceless channel, and it is worth reshooting to get.
Short form and long form
Short
The result and the proof are often the same shot, and the obstacle never gets named. Short form can end at the result: the loop closes and the viewer is satisfied, which is a legitimate outcome rather than a failure.
Long
The result has to be shown twice: once in the hook, and once at the end properly, earned. If the second showing is not better than the first, the viewer feels the video was over at ten seconds. That is the most common structural fault in long-form result videos.
Keeping it honest against the video
The obstacle you name in the hook has to be the obstacle the video actually resolves. A hook that says "we had no plumber" and a video where plumbing never comes up again reads as a lie even when everything else is true, and viewers do not forgive it because it was the specific thing they stayed for. Write the obstacle down before you cut, and check it appears at least twice in the body.
When it goes wrong
The result is shown, and then the video reverses into origin story. "Before I show you how, let me explain how I got into woodworking." The energy of the opening is spent on biography.
The tell: Your retention graph drops between roughly fifteen and forty-five seconds, after a strong start. That is the shape of a hook that was not paid.
The fix: Cut everything between the result and the first real step. Biography goes later, if at all, and usually it turns out not to be needed.
Templates for this pattern
We [did X] every day for [N] days - here’s what happened
I [did X] for [N] days - this is what it did to my [body/life]
Name what it costs to get this wrong. The gap is not curiosity, it is exposure: the viewer suddenly has something to lose.
How it actually works
Stakes work by turning a topic into a personal risk. "How mortgages work" is a topic. "The clause that costs the average buyer eleven thousand" is a risk, and the viewer now cannot comfortably not know.
The stakes have to be theirs, not yours. Your deadline, your budget and your stress are stakes for you and content for them. The question to answer in the hook is what THEY lose by not knowing.
This is the pattern most easily ruined by inflation. Audiences are extremely well calibrated to exaggerated stakes because they see them constantly, and a stake that turns out to be small does more damage than no hook at all: it teaches the viewer to discount everything you say next.
Use it for
Money, health, legal and safety topics
The stakes are genuinely real and you do not have to manufacture anything, which is the only comfortable place to run this pattern.
Buying decisions
A purchase is a fork with a wrong branch, and the viewer is standing at it right now.
Anything with an irreversible step
Irreversibility is stakes in its purest form. Once the wall is down, it is down.
Do not use it for
Low-consequence content
Manufacturing danger around a recipe or a casual vlog reads as manipulation, and it is the fastest way to spend trust you will want later.
Anything you cannot actually resolve
Naming a risk you do not solve leaves the viewer worse off than before they clicked, and they know it.
The part that decides it
What you say next
A stakes hook is paid with RELIEF, and almost nobody does this. Naming a risk raises the viewer's guard: they now feel exposed, and exposure is unpleasant. If the next thirty seconds only escalate, you have built a video people leave to feel better. What you owe them quickly is not the answer, it is the credible existence of one.
0:00 to 0:05
Name the cost in concrete terms. A number, a consequence, a specific thing that happens.
WhyVague risk ("this could really hurt you") is easy to dismiss. A number is not, and it also commits you, which is what makes it feel honest.
0:05 to 0:20
Make it theirs. Say who this happens to, and make it recognisably the person watching.
WhyA risk that belongs to someone else is a story. A risk that belongs to the viewer is a reason to keep watching. This is the beat that decides which one you have.
0:20 to 0:45
Signal the exit. Say plainly that there is a way through and roughly what shape it takes, without giving the whole thing.
WhyThis is the relief. It converts anxiety into attention. Without it, you are asking someone to sit in discomfort on the promise of an eventual fix, and the easier option is always to leave.
The test: Ask whether your first thirty seconds would make a viewer feel worse if they stopped there. If yes, you have raised a risk without offering an exit, and the honest fix is to move the exit earlier rather than to soften the risk.
Hooking, or promising?
Promising means announcing the video instead of delivering. With stakes, it also shows up as describing a topic as important rather than showing what it costs.
PromisingToday we are going to talk about why insurance matters.
HookingThe policy pays nothing if you were on the phone. Mine was in my hand.
The left labels a topic important, which is the weakest possible claim. The right shows the cost and puts the speaker inside it.
PromisingThis is a really important mistake that a lot of people make.
HookingIt is on page four, it is one sentence, and it cost me the deposit.
Importance asserted is easy to ignore. Importance demonstrated with a specific consequence is not.
PromisingYou need to be careful with this, it can go badly wrong.
HookingIf you cut this pipe first, the whole floor floods. Everyone cuts this pipe first.
The left is a warning about nothing in particular. The right names the exact action and tells the viewer they are probably about to do it.
PromisingGetting this wrong can be expensive, so pay attention.
HookingGetting this wrong costs about eleven thousand, and there is a five minute check that catches it.
The right names the cost AND signals the exit in one line, which is the whole stakes pattern compressed into a sentence.
Running it without your face
Workable, and sometimes better faceless, because a stakes hook delivered by an animated face can read as alarmism while the same words over evidence read as documentation.
The document, the clause, the invoice, the scan. Evidence on screen is what stops a stakes hook sounding like a scare.
A number appearing on screen at the moment it is said. It reads as a fact rather than a claim.
A steady, unhurried voice. Urgency in the delivery on top of urgency in the content is what tips this pattern into feeling manipulative.
Before and after of someone who got it wrong, if you have it and can use it fairly.
The risk: Without a face there is nobody to be trusted, so the evidence has to carry the credibility entirely. A faceless stakes hook with no visible proof is indistinguishable from the fear-mongering that fills the same niches, and the audience has already learned to leave.
Short form and long form
Short
The stake and the exit compress into one or two sentences: the cost, then the check that avoids it. Short form is unusually well suited to this pattern, because a single concrete risk with a single concrete fix is a complete unit.
Long
The stake has to be re-raised at least once in the body, or the tension you opened with dissipates around the middle. The usual place is right before the solution, restated with what you now know.
Keeping it honest against the video
The risk you name must be the risk the video resolves, at the same size. A hook that opens on eleven thousand euros and a video that resolves a mild inconvenience has not undersold, it has broken a contract. And if you cannot resolve it, say so in the hook: "there is no clean fix for this, but there is a way to see it coming" is an honest stakes hook and it retains better than a promised solution that never arrives.
When it goes wrong
Escalation with no exit. The first minute stacks consequence on consequence, the tone tightens, and no path is offered until deep into the video, if at all.
The tell: A steep early drop on a topic people genuinely care about. When the subject matters and viewers still leave in the first thirty seconds, the usual cause is that staying felt worse than leaving.
The fix: Move one sentence forward: the one that says a solution exists. It costs you nothing, it does not spoil anything, and it is usually already written somewhere around the two minute mark.
Templates for this pattern
Things I wish I knew before [action]
[N] things I wish I knew before buying [X]
The [N] biggest [category] disasters
The [mistake] that's quietly costing you [money/outcome]
An open loop is a question the viewer now wants answered. Not a preview of what is coming, an actual gap in what they know.
How it actually works
A loop works when the viewer can feel the shape of the answer without having it. They know roughly what kind of thing would resolve this, and they do not have it yet. That gap is what carries them forward.
The most common mistake is confusing a loop with an agenda. "Today I am going to show you three ways to fix your audio" is an agenda: nothing is unresolved, so nothing pulls. "My audio sounded terrible for two years and one setting was doing it" is a loop, because there is now a specific missing fact.
Loops decay. A question left open too long stops feeling like tension and starts feeling like withholding, and the viewer leaves annoyed rather than curious. That is why this pattern lives or dies on what you say next, more than any other.
Use it for
Investigations and explainers with a real answer
There is genuinely something the viewer does not know and you genuinely found it. The loop is not manufactured, which means it will not feel manufactured.
Before and after where the after is not obvious
If nobody can guess how you got there, the how is a loop by itself. If they can guess, it is not, and you are better off with the result pattern.
Anything where you changed your mind
A person reversing a position is an unresolved thing. Viewers stay to find out what moved you.
Do not use it for
Search-intent tutorials
Someone who typed "how to tie a bowline" wants the knot, not suspense. Opening a loop on a search video adds friction between the viewer and the thing they came for, and they will scrub past it or leave. Answer immediately, keep them with quality instead.
Videos whose answer is already in the title
You cannot open a loop the title has closed. If the title gives the number, the hook has to do a different job, usually stakes or result.
The part that decides it
What you say next
This is the part almost nobody teaches, and it is where good openings die. The loop must be PARTIALLY paid fast, roughly within the first fifteen seconds. Not closed, paid. You hand over one concrete piece of the answer so the viewer learns that you actually have it, and then you widen the question.
0:00 to 0:05
Open the loop in one sentence, with a concrete noun in it.
WhyAbstract loops ("something surprising happened") do not catch. A specific object, number or place gives the viewer something to hold while they decide to stay.
0:05 to 0:20
Pay part of it. Give away one real piece of the answer, the piece you would normally save.
WhyThis is the trust transaction. A viewer who receives something concrete this early believes the rest is coming. A viewer who receives a promise instead has no evidence and starts scrubbing. Giving away your best fact early feels wrong and works.
0:20 to 0:45
Widen: show why the partial answer is not the whole story.
WhyThe first payment closes the small loop. If you stop there the video is over at forty seconds. Naming what the partial answer does not explain reopens a bigger question, and that one carries the body of the video.
The test: Read your first fifteen seconds aloud and ask: have I given the viewer a fact they could repeat to someone else? If the honest answer is no, you have written a promise, not a hook.
Hooking, or promising?
One question sorts most weak openings: am I hooking, or am I promising? Promising here means announcing what you are about to do. (That is a different thing from the promise your title makes, which you do want to keep. This is about the sentence that says "I am going to tell you" instead of telling.) The left column below is what people write. The right column is the same information, delivered instead of announced.
PromisingToday I am going to show you three ways to fix your audio.
HookingMy audio sounded terrible for two years, and one setting in the camera menu was doing it.
The left announces an agenda: nothing is unresolved, so nothing pulls. The right hands over a real situation and withholds one specific fact, which is the gap the viewer wants closed.
PromisingIn this video we will look at why this business failed.
HookingThey had nine hundred thousand subscribers and they still could not make rent.
The left describes the video. The right describes the world. Viewers came for the world, and a sentence about the video is a sentence about you.
PromisingStick around, because at the end I will reveal the number.
HookingIt cost four thousand euros. The part I did not expect is where it went.
The left trades a fact for attention and asks the viewer to pay first. The right gives the number away and finds a better question behind it, which costs nothing and buys trust. You almost always have a second question. Use it.
PromisingSomething crazy happened on this trip.
HookingThe taxi driver stopped in the middle of the motorway and got out.
Both are loops, but the left is abstract and the right has a concrete noun in it. Abstract loops do not catch, because there is nothing for the viewer to picture while they decide to stay.
Running it without your face
Better than most patterns, because curiosity travels through voice and image without needing a face to react to. The loop lives in the sentence and in what you choose not to show yet.
The object itself, framed so one part is out of view. What is cropped out IS the loop.
A screen recording that stops one step before the result.
On-screen text carrying the question while b-roll carries the world. Read it aloud too, because a viewer with the sound off and a viewer with the screen in their pocket are both real.
A voice that slows down on the unresolved part. Pacing does the work a face would do.
The risk: Without a face there is no reaction shot to sell that the answer matters, so the delivery has to. Flat narration over a good loop reads as a podcast someone put pictures on. If your read is flat, use the stakes pattern instead: it survives a neutral voice better.
Short form and long form
Short
The loop and its first payment often live in the same sentence, and the whole cycle runs in about three seconds. Short form has no patience for a gap: the viewer is thumb-down, not seated. Practically this means you open on the most specific version of the question and answer part of it before the first cut.
Long
The loop can be wider, and it should be, because it has to survive ten minutes. It still needs partial payment inside fifteen seconds. The difference is not that long form waits longer, it is that long form can afford a bigger question after the first payment.
Keeping it honest against the video
The loop you open has to be the loop the video actually closes. This sounds obvious and is broken constantly: the hook asks one thing, the video is really about a neighbouring thing, and the viewer feels it around the two minute mark without being able to name it. In the retention graph it shows as a drop that starts after the interesting opening, which is usually misread as a pacing problem. It is not. It is the viewer noticing they were answered a different question. Before you cut, write the hook question and the last line of the video next to each other. If the last line does not answer the hook question, one of them is wrong.
When it goes wrong
The loop is opened and then held. "Stay tuned", "we will get to that", "but first, a bit of background". Nothing concrete is handed over in the opening minute.
The tell: You can spot it without any data: if your first sixty seconds could be deleted and the video would still make sense, you were withholding rather than paying.
The fix: Move your strongest single fact into the first fifteen seconds and rebuild the opening around what it does not explain.
Templates for this pattern
[Insider] warns us about [thing you trust]
[Insiders] reveal the [industry] secrets that change how you [do X]
Say the thing the room disagrees with. The gap is the distance between what the viewer believes and what you just claimed.
How it actually works
A contrarian hook works because a claim that contradicts a held belief cannot be filed away. The viewer either wants to see you justify it or wants to see you fail, and both of those keep them watching.
The risk is unique to this pattern: your opening invites argument rather than curiosity. Every second between the claim and your first evidence is a second the viewer spends building a counterargument instead of listening, and once they have built one they are debating rather than watching.
It also spends credibility rather than building it. That is fine if you have some and the claim is true. It is expensive if the claim is really a mild opinion dressed as heresy, which is what most contrarian hooks turn out to be.
Use it for
Genuine expertise against common advice
You have standing, so the claim reads as informed dissent rather than provocation. This is the only fully comfortable use of the pattern.
Things you tried and found wanting
Experience is unarguable in a way opinion is not. "I did this for two years and it did not work" cannot be countered by someone who has not.
Widely repeated advice with a real exception
The advice is usually right, and the video is about the case where it is not. Precise disagreement is far stronger than blanket disagreement.
Do not use it for
Topics where you have no standing
A contrarian claim from nowhere is just noise, and the comment section becomes the content while the video becomes the pretext.
Claims you soften later
If the video walks the claim back to something reasonable, the hook was a bait and the viewer notices at exactly the moment you were hoping they would agree with you.
The part that decides it
What you say next
A contrarian hook is paid with EVIDENCE, and the window is less forgiving than anywhere else in this book. Not the shortest, the most hostile: you have started an argument, and you are not in the room to defend it. What you say next has to give the viewer a reason to suspend the counterargument they are already assembling, and it has to arrive before they finish assembling it.
0:00 to 0:05
State the claim flatly, once, with no hedging.
WhyHedged heresy is the worst of both: it annoys the people who agree and fails to interest the people who do not. If you are going to say it, say it.
0:05 to 0:20
Establish standing or give the first piece of evidence, whichever is stronger.
WhyThis is the beat that decides whether the rest is watched or argued with. Standing is faster ("I have shipped forty of these"), evidence is stronger. If you have neither, do not run this pattern.
0:20 to 0:45
Steelman the thing you are attacking. Say why the common view exists and when it is right.
WhyThis is what separates a useful contrarian video from a hot take, and it is the beat almost everyone skips. Granting the opposing case makes your disagreement precise, and precise disagreement is much harder to dismiss than blanket disagreement.
The test: Ask someone who holds the view you are attacking to watch the first forty five seconds. If their response is an argument, you have not paid. If it is "go on", you have.
Hooking, or promising?
Promising means announcing rather than delivering. With contrarian hooks, promising is the hedge: signalling that a disagreement is coming instead of stating it.
PromisingI think there are some problems with the advice everyone gives about this.
HookingPosting daily is why most small channels stall.
The left announces that a disagreement exists. The right IS the disagreement. Announcing a claim you are about to make is the most common way this pattern gets defused.
PromisingThis might be a bit of an unpopular opinion, but hear me out.
HookingNinety percent of the gear in that recommended list is a waste of money.
The preamble asks permission and lowers the stakes of your own claim before you make it. Say the thing; the audience will decide whether it is unpopular.
PromisingA lot of people are wrong about this and I want to explain why.
HookingThe rule is right, and following it is what is holding you back. Both of those are true.
The left is a claim about other people. The right is a claim about the world, and it has enough shape that the viewer can already sense the argument.
PromisingIn this video I will challenge the conventional wisdom on pricing.
HookingI doubled my prices and lost one client. He was the problem.
The left describes the video. The right supplies the claim and the first evidence in the same breath, which is exactly what this pattern needs fastest.
Running it without your face
The hardest pattern to run faceless, because a face carries standing for free and you do not have one. It is workable, but the burden shifts entirely onto evidence and it has to arrive earlier.
Data on screen within the first ten seconds. Numbers substitute for a person the audience can trust.
Receipts: the invoice, the analytics screen, the timestamped record. Evidence you could not have fabricated does the job standing would have done.
Citing the thing you disagree with, verbatim and fairly. Quoting the opposing case accurately is itself a credibility signal.
A calm read. Faceless plus heated equals a rant, and rants are argued with rather than watched.
The risk: Standing has to be stated rather than embodied, and stated standing sounds like boasting unless it is specific and instrumental. "I have shipped forty of these" works. "As an expert in this field" does not, and it is worse than saying nothing.
Short form and long form
Short
Claim plus one piece of evidence, and stop. There is no room to steelman, which means short form contrarian content is inherently less fair than long form. That is worth being uncomfortable about rather than treating as a format quirk.
Long
The steelman is mandatory, and it should be longer than feels comfortable. The most persuasive contrarian long-form videos spend a genuine chunk making the opposing case properly before dismantling it, and the ones that skip it read as one-sided even to viewers who agree.
Keeping it honest against the video
The claim you open with has to be the claim you defend, at full strength, all the way through. The common failure is a hook stated at ten and a video that argues for six, landing on a reasonable position nobody would have disagreed with. That is not nuance arriving, that is the hook having been an exaggeration. If your honest position is a six, open at six: it is a smaller hook and a much better video.
When it goes wrong
The claim is made, and then the video spends two minutes on context, definitions, or a story about how the creator came to this view, while the viewer waits to find out whether the claim is serious.
The tell: The comments argue with the hook rather than the video, and they repeat objections you actually answered. That means the objection was formed before your answer arrived.
The fix: Move your single strongest piece of evidence to within twenty seconds of the claim, even if it breaks a chronology you liked. Chronology is worth less than being believed.
Show something they did not expect to see. The gap is confusion, and confusion is the shortest-lived of all the gaps.
How it actually works
A pattern break works on the feed rather than on the topic. Everything around your video looks like what that niche looks like, and yours does not, so the scroll stops. It buys attention that has nothing to do with your subject.
That is also its weakness. Attention borrowed from surprise has to be converted into attention about the video, and the conversion window is very short. Confusion that is not resolved becomes irritation faster than any other opening state.
The single most important rule: the break must come FROM the video. A striking image unrelated to your content is not a pattern break, it is bait, and the viewer works this out within about twenty seconds and leaves with a slightly worse opinion of you.
Use it for
Videos with a genuinely strange moment in them
You are not manufacturing anything, just leading with the oddest true thing you have. This is the pattern at its best.
Familiar topics in saturated niches
When every thumbnail and opening in your category looks the same, being visually different is worth more than being marginally better.
Anything with a strong physical or visual anomaly
Visual anomalies survive muted autoplay, which is how a large share of your audience meets the video.
Do not use it for
Breaks unconnected to the content
It is bait, it is detected quickly, and the cost is not one video: it is the viewer discounting your next opening too.
A channel that does it every time
A pattern break that is your pattern is not a break. Once the audience expects the strange opening, it has become your format and it stops buying anything.
The part that decides it
What you say next
A pattern break is paid with ORIENTATION, and the clock is shorter than anywhere else. The viewer is holding a question they cannot even phrase: what am I looking at. Until that is answered they cannot become interested in the video, because they are still processing the image. Everything else waits.
0:00 to 0:03
The break. Hold it long enough to be seen properly, which is longer than feels right in the edit.
WhyA break cut too fast reads as a glitch. The viewer needs enough frames to register the anomaly, and a beat too long is much cheaper than a beat too short here.
0:03 to 0:15
Orient. Say or show what it is, where you are, and why this is happening.
WhyThis is the payment. It resolves the confusion and converts a stopped scroll into a watching viewer. Skipping it is the single most common way a great opening shot wastes itself.
0:15 to 0:40
Connect it to the video. Make the strange thing the reason the video exists, or the question it answers.
WhyOrientation alone leaves the viewer satisfied and free to go. Connection turns the anomaly into a stake in the rest of the video, which is what you actually wanted from it.
The test: Show the first fifteen seconds to someone with no context and ask what the video is about. If they can describe the image but not the video, you have oriented them and not connected, and the drop will come around thirty seconds.
Hooking, or promising?
Promising means announcing rather than delivering. With pattern breaks, the failure is usually the opposite of announcing: showing something odd and then explaining the video instead of the image.
PromisingYou are probably wondering what this is, so let me back up and explain how we got here.
HookingThat is a hotel corridor in a decommissioned power station. It is eleven at night and we are not supposed to be here.
The left announces an explanation. The right IS the explanation, and it takes less time than announcing it would have.
PromisingBefore I explain this shot, a quick word about what this channel does.
HookingThe wall is moving because the whole building is on hydraulics.
Channel context is never the payment for a pattern break. The image raised the question; answer that one.
PromisingCrazy start to the video there, anyway, welcome back everyone.
HookingIt came off in my hand. That is the part that was holding the mast on.
The left treats the break as a decorative opener and then resets to a normal intro, which wastes it entirely. The right treats it as information and keeps going.
PromisingThis is going to be a very different kind of video for me.
HookingI have not spoken to a person in nine days. That was the experiment.
Announcing difference is not being different. The right states the strange fact plainly and lets it be as odd as it is.
Running it without your face
The most faceless-friendly pattern of the five. A pattern break is an image, images do not need a presenter, and the orientation beat can be carried by a caption as easily as a voice.
The anomaly, held. The whole hook is one shot chosen well and not cut early.
A caption that orients in under eight words. Faceless orientation is often faster in text than in speech.
Sound design doing the work of a reaction. The right sound tells the viewer how to feel about what they are seeing without anyone in frame.
A second angle of the same thing. Confirming the anomaly from another position is the fastest way to say "yes, this is real".
The risk: Orientation is easy to forget when there is no presenter whose job it obviously is. Faceless pattern breaks fail most often by staying beautiful and unexplained for forty seconds, which is a mood piece rather than a video opening.
Short form and long form
Short
The break and the orientation are frequently the entire video, and that is fine: a strange thing shown and explained is a complete short. Connection to a larger point is optional here in a way it never is in long form.
Long
Orientation must arrive inside fifteen seconds and connection inside forty, or the opening becomes a thing that happened rather than a reason to watch. Long form is where unconnected pattern breaks do their real damage, because there are ten more minutes for the viewer to notice the opening did not matter.
Keeping it honest against the video
The break has to earn its place in the finished video, not just in the opening. If the strange thing never returns, the viewer registers it as decoration, and decoration at the top of a video teaches them that your openings are not load-bearing. The strongest version brings it back at the end with more understanding than they had at the start, which converts the opening from a trick into a frame.
When it goes wrong
A striking opening shot, then a hard reset into a normal intro: channel branding, a greeting, a preamble. The anomaly is never mentioned again.
The tell: Strong first three seconds in the retention graph and a cliff before thirty. That specific shape almost always means an unpaid pattern break rather than a pacing problem.
The fix: Delete the reset. Go straight from the image to what it is, and put the greeting somewhere it is not competing with an unanswered question, which usually means nowhere.
If you want to be walked through writing a hook from scratch, the
free hooks module does that across five lessons. If you want a
formula to fill in right now, the hook library has 58 of them
sorted by pattern and niche. This book is the layer underneath both: why each pattern behaves
the way it does, and what to do with the seconds after it.