A hook is a contract. It opens a gap and the rest of the post has to close it. That is the whole mechanism, and it explains why the strongest-sounding openings sometimes perform worst a promise that goes unpaid costs more than a promise never made.
The 50 below are grouped by the mechanism each one uses rather than by mood, because templates wear out and mechanisms do not. Every entry carries the condition under which it backfires, which is usually the more useful half.
The three-second window
The common framing is that a hook exists to stop the scroll. That is only half right. Stopping is cheap. A bright frame or a loud first syllable can do it and stopping alone does nothing useful.
What the platforms respond to is whether people stayed. A post that stops thousands of thumbs and then loses them at second three reads, from the outside, as content that did not deliver. The hook’s real job is narrower than it sounds: survive the swipe, then hand over to content good enough to keep the viewer.
This reframes what a good opening looks like. It is not the most dramatic line available. It is the most accurate interesting line available the one that describes what is genuinely coming.
A hook does not buy views. It buys the content a fair hearing.
Three channels, not one line
On Reels and TikTok a viewer receives three things at once in the opening moment: the first frame, the first spoken words, and whatever text sits on screen. Most hook advice only supplies the second of those.
Treating them as three separate jobs is where a lot of quick improvement lives, because each one has a distinct failure mode.
The on-screen text matters more than it looks. A large share of Reels views arrive muted, and text that simply repeats the spoken line wastes the one channel that still works in silence. Text should say something the audio does not, so a silent viewer and a listening viewer each receive a complete opening.
The first frame is where most posts quietly lose. Opening on a static title card, a brand logo, or a person settling into position spends the most valuable moment on setup. Opening mid-gesture, mid-sentence or mid-process spends it on motion.
Hook debt
When a hook promises more than the content delivers, the gap between the two is worth naming. Call it hook debt.
The pattern is familiar. An opening claims something startling, the viewer stays for four seconds, realises the payoff is ordinary, and leaves. Retention collapses, reach follows, and the conclusion drawn is usually that the hook failed.
It did not. The hook worked and it delivered attention the content could not service. The fix runs in the opposite direction to instinct: not a bigger promise, but a smaller and more precise one.
Over-promised: “This changed everything about how I cook.”
What the video shows: a better way to dry herbs.
Rewritten downward: “I stopped throwing out half my herbs when I changed one thing.”
The second version is less exciting and performs better, because everyone who stays for it gets exactly what they came for. Debt is repaid at the moment the promise is made, not later. A strong opening also needs the rest of the Reel script to carry that promise through the setup and payoff.
Where the hook lives on each surface
The same idea needs a different carrier depending on where it is published. These are the mechanical differences, not stylistic preference.
| Surface | The hook lives in | Main constraint |
| TikTok | First words, supported by the frame | Sound is usually on; talking-head cold opens are well tolerated |
| Reels | First frame and on-screen text | Large muted share, and often served to people with no context for the account |
| Carousels | The cover slide | Static the hook has to work as a still image with no motion and no sound |
| Text posts | The line above the fold | Competes with a truncation point rather than a swipe |
| Shorts | First frame plus title | Viewers arrive scrolling but stay longer once a premise is clear |
Two practical consequences follow. On carousels, motion-based hooks are unavailable, so the cover has to carry a complete promise in one readable line. On text posts, the hook competes with a “see more” cut, which means the interesting word needs to arrive before the break rather than at the end of a sentence.
The 50 Hook Ideas
Each entry gives the template, the mechanism it runs on, a filled example, and the condition where it stops working. The bracketed slots are the parts to replace.
1. Curiosity gap (1–7)
These open a loop the brain wants closed. The loop has to be closable within the length of the post, which is the constraint most of them fail on.
01 “Nobody warned me about [specific thing] until [too-late moment].”
MECHANISM withheld information plus an implied cost already paid.
FILLED “Nobody warned me about clause 14 until the lease was signed.”
FAILS WHEN the thing is generic. “Nobody warned me about adult life” closes no loop because it opens none.
02 “There’s a reason [common frustration] keeps happening.”
MECHANISM promises causation for something the viewer has already experienced.
FILLED “There’s a reason your sourdough keeps coming out flat.”
FAILS WHEN the reason turns out to be the obvious one everybody already suspected.
03 Open on the finished result, then cut back to the start.
MECHANISM reverse chronology; the outcome creates demand for the process.
FILLED a restored chair in full light for one second, then the wreck it came from.
FAILS WHEN the result is not striking enough to justify rewinding. Weak outcome, no demand.
04 “[Number] things happen when you [action]. The [nth] one is the problem.”
MECHANISM an enumerated loop with one item flagged in advance, which sets a destination.
FILLED “Three things happen when you air-dry hair. The third one is the problem.”
FAILS WHEN the flagged item is the weakest of the set. The flag raises expectations on that one item specifically.
05 “This took [long time] to work out. It takes 40 seconds to explain.”
MECHANISM effort asymmetry; the viewer is offered a shortcut past someone else’s cost.
FILLED “This took two years of failed batches. It takes 40 seconds to explain.”
FAILS WHEN the explanation runs well past the time stated. The number is a promise, not a figure of speech.
06 “I found [unexpected thing] in [unlikely place].”
MECHANISM anomaly. Something is where it should not be, and the mind wants the explanation.
FILLED “I found a 1962 receipt inside the piano.”
FAILS WHEN the explanation is guessable from the hook itself, which closes the loop before the content can.
07 “I almost didn’t post this.”
MECHANISM implied risk. Hesitation signals that the content carries some cost to share.
FILLED “I almost didn’t post this, because it makes me look bad.”
FAILS WHEN nothing risky follows. This one burns audience trust faster than almost any other, and it does not recover.
2. Contradiction and pattern interrupt (8–14)
These collide with something the viewer already believes. The collision has to be real; a contradiction that resolves into conventional advice reads as a bait.
08 “[Widely repeated advice] is wrong.”
MECHANISM direct collision with an accepted belief, which demands resolution.
FILLED “Drinking eight glasses of water a day is not a real rule.”
FAILS WHEN no evidence arrives in the next few seconds. Contradiction without support reads as noise.
09 “Stop [common action].”
MECHANISM imperative reversal. An instruction to stop implies the viewer is currently doing harm.
FILLED “Stop rinsing pasta.”
FAILS WHEN most of the audience never did it. The hook lands on nobody.
10 “The more I [effort], the worse [outcome] got.”
MECHANISM inverse relationship. Effort producing decline is counter-intuitive enough to demand a reason.
FILLED “The more I watered it, the faster it died.”
FAILS WHEN the inversion is well known in that niche, where it becomes a truism rather than a surprise.
11 “[Expensive option] and [cheap option] do the same job.”
MECHANISM value collapse. Price is a proxy for quality, and breaking it is immediately interesting.
FILLED “The £4 version and the £90 version are the same tool.”
FAILS WHEN they are not actually the same, and the difference is visible in the content itself.
12 “Everyone recommends [X]. Almost nobody does [X].”
MECHANISM exposes a gap between stated and actual behaviour, which invites self-examination.
FILLED “Everyone recommends warming up. Watch any gym at 6pm.”
FAILS WHEN the gap is presented as hypocrisy rather than explained, which reads as scolding.
13 “I did [thing] for [duration] and nothing happened.”
MECHANISM anti-result. The genre trains viewers to expect transformation, so its absence is the surprise.
FILLED “I ran every morning for 60 days and my times got worse.”
FAILS WHEN the content then pivots to a transformation anyway. The hook has to stay honest to work twice.
14 “This is the best [thing] I’ve used and I’d never buy it again.”
MECHANISM internal contradiction in a single sentence, which cannot be resolved without watching.
FILLED “Best camera I’ve ever owned. Never buying another one.”
FAILS WHEN the resolution is trivial, such as simply being expensive. The contradiction needs a real reason.
3. Stakes and loss (15–20)
Avoiding a loss tends to move people more reliably than pursuing an equivalent gain. These hooks lean on that, which also makes them the easiest group to overplay.
15 “[Habit] is costing [specific amount] and most people don’t notice.”
MECHANISM quantified, ongoing loss. A number makes an abstract waste feel concrete.
FILLED “That standby light is about £30 a year and nobody checks.”
FAILS WHEN the figure is too small to matter, at which point the hook argues against itself.
16 “If [condition], stop before you [action].”
MECHANISM a warning gate. It filters for a specific group and gives them a reason to pause.
FILLED “If the pan is non-stick, stop before you preheat it empty.”
FAILS WHEN the condition applies to almost nobody watching, which wastes the filter.
17 “This mistake cost me [specific loss].”
MECHANISM personal stake. Someone has already paid, so the lesson arrives pre-tested.
FILLED “This mistake cost me a full week of footage.”
FAILS WHEN the loss is vague. “Cost me a lot” carries none of the weight of a named number or object.
18 “There’s [timeframe] to [action] before [consequence].”
MECHANISM deadline pressure, which converts interest into immediate attention.
FILLED “There are about 20 minutes to move a cutting before the stem seals.”
FAILS WHEN the deadline is invented. Audiences recognise manufactured urgency quickly and it costs credibility across future posts.
19 “Check [thing] before [routine event].”
MECHANISM pre-emptive inspection. It attaches to something the viewer is already going to do.
FILLED “Check the expiry on the extinguisher before the clocks change.”
FAILS WHEN the routine event is not shared by the audience, breaking the attachment.
20 “Once [event] happens, [outcome] can’t be undone.”
MECHANISM irreversibility. Permanence raises the value of knowing in advance.
FILLED “Once it’s been through a hot wash, the shape doesn’t come back.”
FAILS WHEN it is in fact reversible, and a comment section points that out within the hour.
4. Identity and in-group (21–26)
These work by telling a viewer that the post is about people like them. The precision of the group definition does all the work a broad group flags nobody.
21 “If [very specific behaviour], this one’s for you.”
MECHANISM self-selection. The viewer identifies themselves rather than being told.
FILLED “If there’s a drawer of cables nobody has opened since 2019, this one’s for you.”
FAILS WHEN the behaviour is universal. “If you’re tired” selects everyone, which is the same as selecting no one.
22 “[Specific group] will get this immediately.”
MECHANISM in-group flag. Insiders feel recognised; outsiders stay to find out what they are missing.
FILLED “Anyone who has worked a closing shift will get this immediately.”
FAILS WHEN the content then explains the reference at length, which insults the group it just flattered.
23 “There are two kinds of [group].”
MECHANISM taxonomy invitation. Viewers begin sorting themselves before the sentence finishes.
FILLED “There are two kinds of people at a buffet.”
FAILS WHEN the two categories are not genuinely distinct, leaving nothing to sort into.
24 “This is what [role] actually do all day.”
MECHANISM insider access. The word “actually” implies the public version is incomplete.
FILLED “This is what a court interpreter actually does all day.”
FAILS WHEN the reality matches the assumption, so the promised gap never appears.
25 “[Place] people explain this.”
MECHANISM local call plus an open question, which does double duty as a comment prompt.
FILLED “Glasgow people, explain this bus stop to me.”
FAILS WHEN the thing is not actually regional, and the local framing looks arbitrary.
26 “It takes about [short time] to tell whether someone [trait].”
MECHANISM a judgement claim, which prompts viewers to test themselves against it.
FILLED “It takes about ten seconds to tell whether someone has actually cooked before.”
FAILS WHEN the tell is unfalsifiable, which turns an interesting claim into an empty one.
5. Specificity (27–32)
Precise figures read as evidence of real work. Rounded ones read as estimates. This group needs no cleverness at all, only accuracy.
27 “I went through [precise number] of [thing]. [Small number] of them [pattern].”
MECHANISM implied labour. A large sample suggests the conclusion was earned.
FILLED “I read 340 rejection emails. Four sentences turned up in almost all of them.”
FAILS WHEN the number is rounded to a marketing figure. 340 is credible; 500 is not.
28 “It costs exactly [precise figure].”
MECHANISM precision as proof. An exact cost implies someone actually checked.
FILLED “Start to finish, it came to £11.40.”
FAILS WHEN major costs are excluded from the figure, which the comments will find.
29 “Day [n] of [ongoing project].”
MECHANISM serialisation. A counter implies both history and a future instalment.
FILLED “Day 61 of restoring a barn nobody wanted.”
FAILS WHEN each instalment looks identical to the last, at which point the counter measures repetition rather than progress.
30 “[Number] minutes, [number] ingredients, [number] steps.”
MECHANISM constraint stack. Three limits together signal that the whole thing is manageable.
FILLED “Nine minutes, four ingredients, one pan.”
FAILS WHEN the constraints are met only by excluding obvious steps, which reads as a technicality.
31 “On [date] at [time], [event].”
MECHANISM timestamp. A specific moment frames the content as documentation rather than opinion.
FILLED “On 3 March at 4:40am, the alarm went off for the fourth night.”
FAILS WHEN the date carries no significance, making the precision decorative.
32 “Out of [large number], only [small number] [outcome].”
MECHANISM ratio. A stark fraction is easier to feel than a percentage.
FILLED “Out of 90 applicants, two got a reply.”
FAILS WHEN the ratio is unremarkable for that field, and specialists in the comments say so.
6. Story cold open (33–38)
These begin mid-scene and let the viewer assemble the context while watching. The discipline is refusing to supply setup the moment setup appears, the tension goes.
33 “So [person] rings me at [odd hour] and says”
MECHANISM in medias res. Starting mid-event makes the viewer chase the context.
FILLED “So my landlord rings me at 11:40 on a Sunday and says”
FAILS WHEN the next line is thirty seconds of background, which undoes the cold open entirely.
34 “I wasn’t supposed to be there.”
MECHANISM displacement. Someone in the wrong place implies both a story and a consequence.
FILLED “I wasn’t supposed to be in that meeting at all.”
FAILS WHEN being there turns out to be unremarkable, leaving the opening unexplained.
35 “The email said [short quoted fragment].”
MECHANISM document open. A quoted artefact is concrete in a way that summary is not.
FILLED “The email said ‘please disregard the previous email.’”
FAILS WHEN the fragment is too long to absorb in one beat. Short enough to read at a glance, or it is not a hook.
36 “[Short time] before [known event], [complication].”
MECHANISM countdown frame. A fixed deadline plus a problem creates tension immediately.
FILLED “Three days before the wedding, the venue called.”
FAILS WHEN the complication resolves easily, which makes the countdown feel like set dressing.
37 “I kept [object] for [long period]. Here’s why.”
MECHANISM unexplained attachment. Duration implies significance not yet revealed.
FILLED “I kept a broken watch for eleven years. Here’s why.”
FAILS WHEN the reason is sentimental in a way the viewer could have guessed from the object alone.
38 “It started as [small, harmless thing].”
MECHANISM escalation promise. “Started as” guarantees the scale changed.
FILLED “It started as one shelf.”
FAILS WHEN the escalation is mild. The gap between start and end is the entire payoff.
7. Direct address (39–44)
Being singled out is hard to scroll past. This group is also the easiest to overuse, and an account that opens every post this way trains its audience to ignore it.
39 “If this reached you, it’s because [inferred situation].”
MECHANISM algorithmic address. It treats the recommendation itself as evidence about the viewer.
FILLED “If this reached you, something’s been rattling in the car for a while.”
FAILS WHEN the inference is wrong for most viewers, which makes the account look like it is guessing.
40 “Pause here.”
MECHANISM interrupt command. A direct instruction breaks scrolling momentum.
FILLED “Pause here look at the corner of the frame.”
FAILS WHEN there is nothing worth pausing on, which makes the next instruction from that account easy to ignore.
41 “Do this now, before it goes out of your head.”
MECHANISM immediate action. It converts watching into doing while attention is still held.
FILLED “Do this now open settings and check the backup date.”
FAILS WHEN the action takes longer than a moment, so nobody does it and the framing is wasted.
42 “Tell me I’m wrong about this.”
MECHANISM invited disagreement, which turns viewers into participants.
FILLED “Tell me I’m wrong: most home offices don’t need a second monitor.”
FAILS WHEN the position is one nobody disputes, leaving the invitation unanswered.
43 “This is the part everyone skips.”
MECHANISM meta-address. Naming the skip makes skipping feel like a deliberate choice.
FILLED “This is the step everyone skips, and it’s why the paint lifts.”
FAILS WHEN the part is genuinely skippable, which proves the viewer right for skipping it.
44 “Two options. Pick one before the end.”
MECHANISM forced choice with a deadline, which gives a reason to watch through.
FILLED “Two ways to lay this floor. Pick one before the end.”
FAILS WHEN one option is obviously better, which removes the decision.
8. Visual and silent hooks (45–50)
These use no spoken line at all. They matter most on Reels, where a large share of playback starts muted, and on carousels, where sound does not exist.
45 Open mid-action, with hands already moving.
MECHANISM motion implies a process already underway, so the viewer joins rather than waits.
FILLED first frame is a knife already through the onion, not a board being arranged.
FAILS WHEN the action is unrecognisable out of context, and the frame reads as visual noise.
46 Put one unexplained object in the frame.
MECHANISM visual anomaly. Something that does not belong holds attention while the mind works on it.
FILLED a bike wheel on a kitchen counter, never mentioned for the first few seconds.
FAILS WHEN the object is never explained at all, which reads as a trick rather than a setup.
47 Show the finished result as frame one.
MECHANISM outcome-first. The result functions as the promise, with no words needed.
FILLED the plated dish in full light before a single ingredient appears.
FAILS WHEN the result is ordinary. This hook is only as strong as the thing it shows.
48 Let something enter or leave the frame unexpectedly.
MECHANISM physical pattern interrupt. Sudden change in a static frame is difficult to ignore.
FILLED a full shelf, then one item tipping out of shot in the first half-second.
FAILS WHEN the interrupt is unrelated to the content, which feels like a bait once the subject arrives.
49 Text alone on a plain frame.
MECHANISM the absence of visual competition makes a single line unusually readable.
FILLED one sentence, centred, on a still background, held for a beat before the video starts.
FAILS WHEN the line runs long. A plain frame gives a sentence nowhere to hide.
50 Open on a reaction to something off-screen.
MECHANISM displaced attention. A visible response to an invisible cause creates an immediate question.
FILLED a face registering something the camera has not shown yet.
FAILS WHEN the reveal is slower than a few seconds, at which point the question goes stale.
Turning a template into a line
A template is not a hook. It becomes one when the bracketed slot is filled with the most specific thing available, and specificity is where most adaptations stall. The same principle applies to broader hook formulas: keep the structure, but rebuild the wording around the topic and audience.
A method that works reliably:
• Find the variable slot and fill it with the most concrete version. Not “a lot of money” but “£340”. Not “an old tool” but “a 1978 hand drill”. Concrete detail is what makes a borrowed template stop sounding borrowed.
• Read the result aloud and cut everything before the interesting word. Most drafts open with two or three words of throat-clearing, and deleting up to the first noun that carries meaning usually improves the line.
• Check it still makes sense to someone who has never seen the account. Posts reach strangers constantly, and a hook that depends on prior context loses everyone who lacks it.
• Try the same content through three different mechanisms before committing. The first hook that comes to mind is usually the most obvious one, which means it is also the one most other accounts arrived at.
Running one topic through several groups shows how differently the same material can open. Take a post about a cheap repair to a leaking tap:
Specificity (#28): “The part cost £2.40.”
Contradiction (#11): “The plumber quoted £120 for a washer.”
Stakes (#17): “I ignored this drip for four months. It cost me a cabinet.”
Visual (#47): Open on the dry, finished tap, then cut to the leak.
Same content, four openings, four different audiences. The right choice depends on which one the content can actually pay for.
Reading the retention graph
Comparing hooks fairly requires holding everything else steady same content, similar posting window, same thumbnail. Two posts with different hooks and different topics tell us nothing.
The retention curve is more informative than the view count, because the shape of it points at the cause.
| Curve shape | Likely cause | What to change |
| Steep cliff in the first seconds | The opening did not earn the next beat | The hook itself, or the first frame |
| Drop shortly after a strong start | Hook debt the promise outran the payoff | Lower the promise or raise the payoff |
| Gradual decline through the middle | Pacing, not the hook | Trim the middle; the opening is fine |
| Healthy retention, low reach | Distribution, not the hook | Leave the hook alone and look elsewhere |
That last row is the one most worth internalising. A post can have an excellent hook and still travel nowhere, and rewriting the opening in response will only make a working thing worse.
Why hooks stop working
Every successful template eventually saturates. It spreads, gets imitated, becomes recognisable as a format rather than a sentence, and viewers start scrolling on sight.
The reliable tell is that a format has reached brand accounts and paid advertising. By the time a phrasing turns up in an ad, its useful life as an organic hook is largely over.
This is why the grouping above matters more than the fifty lines themselves. Templates saturate; mechanisms do not. Curiosity, loss aversion and in-group recognition have not worn out in the entire history of persuasion, and they are not going to wear out on a short-video feed.
Rotate mechanisms, not phrasings. The phrasings expire; what sits underneath them does not.
In practice that means treating this list as a set of eight tools rather than fifty lines. When a particular wording starts underperforming, the move is to keep the mechanism and rebuild the sentence.
The pre-publish scorecard
Six checks, each answerable in a few seconds.
• The payoff is named or implied within the first three seconds. Not the full answer, but enough that a viewer knows what they are waiting for.
• It works with the sound off. On-screen text should carry the premise on its own, without simply transcribing the audio.
• There is no throat-clearing. Greetings, introductions and “so basically” all spend the most valuable moment on nothing.
• The promise matches what the content delivers. If the opening is more exciting than the post, the opening needs lowering rather than the post inflating.
• It makes sense to a stranger. Assume no prior knowledge of the account, the series, or any running joke.
• The first frame is not a title card or a logo. Motion, a face, or an object in use will outperform a static introduction almost every time.
The verdict
Hooks get discussed as though they were the hard part, and they are not. Writing an arresting opening is comparatively easy; writing one the content can pay for is the actual constraint.
Which is why the fifty above are organised by mechanism and paired with failure conditions. A list of lines runs out the moment the lines saturate. A set of mechanisms keeps generating new lines for as long as the underlying content has something worth opening with.
Comments