Hooks for short videos: 20 formulas in the first 3 seconds that hold attention

70% of viewers swipe a short video within the first 3 seconds. Not in 10 - in 3. You don’t have a minute to “accelerate”, there is no time to explain the context, there is no right to a boring introduction. The hook - the first 2-3 seconds of the video - decides everything: they will watch it or scroll through it, it will get into recommendations or die after 200 views. For an affiliate who feeds through a network of accounts, the hook determines ROI of the entire bunch. This article contains 20 proven hook formulas, organized into five categories, with specific examples for TikTok, Reels, and Shorts. Each formula can be adapted to any vertical - nutra, gambling, dating, merchandise - and tested on real traffic.

Why the first 3 seconds decide everything

Before you analyze the formulas, you need to understand what you are fighting against. The TikTok, Instagram Reels and YouTube Shorts algorithms work on the same principle: the video receives a “test” reach of 200–500 people, and if the retention metrics are good, the video is pushed further. If not, he dies.

Key metric is retention rate at the 3-second mark. This is the percentage of viewers who did not swipe the video in the first 3 seconds. Average performance by platform in 2026:

• TikTok: average retention for 3 seconds - 32%. Videos included in recommendations show 55%+ at this mark
• Instagram Reels: average - 28%. To get into Explore you need 50%+
• YouTube Shorts: average - 35%. To get into the feed Shorts - 52%+

What does this mean in practice? Two-thirds of viewers leave without watching until the fourth second. Your ideal offer, your well-developed script, your selling CTA - all this is useless if the hook doesn’t catch you in the first moments. Content can be brilliant from the 4th second - but no one will see it.

For arbitrage traders this is amplified multiplicatively. If you have a network of 30 accounts and each video gets 200 views instead of 20,000 - because the hook is weak - you lose not just one video, but the entire economy of the bundle. The hook is a lever with maximum leverage: improving retention from 30% to 55% can increase reach by 10–50 times due to algorithmic pushing.

Another critical point: platforms evaluate retention of the first seconds as the main signal of content quality. TikTok in 2026 uses early retention (0-3 seconds) as a key factor in the decision to expand coverage. Instagram Reels looks at what percentage of viewers “stopped scrolling” - that is, they stayed on the video for more than 2 seconds. YouTube Shorts analyzes the “swipe-away rate” - the percentage of immediate swipes. All three platforms essentially ask the same thing: “Did the first few seconds hook the viewer enough to stay?”

Category 1: Shock and curiosity - 4 formulas

The most aggressive category of hooks. It works due to cognitive dissonance: the viewer sees or hears something that contradicts his expectations, and the brain demands that this contradiction be resolved. It's impossible to scroll through until you find out what's going on. On TikTok, shock hooks average 58-65% retention at the 3-second mark - the best result among all categories.

Formula 1: “You’re doing it wrong”

Mechanics: directly blaming the viewer for the mistake. Causes an instant reaction - “How wrong is this? I know exactly what I’m doing!” The viewer stays to defend his position or to see that he is wrong.

Examples:

• “You apply the cream to your face incorrectly - and that’s why it doesn’t work” (guts)
• “You bet incorrectly - that’s why you constantly lose” (betting)
• “You fill out the form incorrectly and miss 90% of matches” (dating)

Why it works: Dunning-Kruger effect in reverse. Nobody believes that they are doing basic things wrong. The phrase causes slight irritation and a need to be convinced - and the viewer looks further. Retention at 3 seconds - 60–67%.

Formula 2: “99% of people don’t know about this”

Mechanics: exclusive information. The viewer wants to be in the 1% of those who know. FOMO (fear of missing out) works - fear of missing out on important information.

Examples:

• “99% of people don’t know this way to remove wrinkles in a week” (guts)
• “99% of players don’t know this strategy - it gives a plus every month” (gambling)
• “99% of arbitrage traders do not use this combination - and lose half of their profit”

Why it works: exclusivity trigger + social proof (“if 99% don’t know, this is really valuable information”). You can vary the percentage: 97%, 95%, “almost no one.” Retention — 55–62%.

Formula 3: “Stop [the action] right now”

Mechanics: ultimate requirement. The phrase carries a threat - if you don’t stop, there will be consequences. The viewer lingers to understand what the consequences are and why he needs to stop.

Examples:

• "Stop washing your face in the morning - and here's why" (guts)
• “Stop placing express bets - you’re just giving money to the bookmaker” (betting)
• "Stop pouring on one account - it's killing your ROI"

Why it works: imperative creates a sense of urgency. The viewer thinks, “I do this every day—what am I missing?” Negative wording is more compelling than positive wording. Retention — 57–63%.

Formula 4: “This should be prohibited”

Mechanics: a hint of something so effective that it is on the verge of being banned. The brain processes this as “secret knowledge that is hidden.”

Examples:

• “This cream is so effective that it should be banned” (guts)
• “This strategy should be banned - bookmakers hate it” (betting)
• "This software should be banned - it makes it too easy to scale traffic"

Why it works: “forbidden fruit” is an ancient psychological trigger. If something “should be banned,” it means it works so well that it interferes with someone powerful. Retention — 56–61%.

Category 2: Question hooks - 4 formulas

A question causes the brain to automatically search for an answer - this is a reflex that cannot be turned off. While the viewer is formulating an answer, he is already watching your video. Question hooks are especially effective in Reels - older audiences value intellectual engagement over shock value.

Formula 5: “Did you know that...?”

Mechanics: A classic hook question with a surprising fact. The goal is to find a fact that the viewer really doesn’t know (or thinks he doesn’t know). A trivial fact will kill a video faster than the absence of a hook.

Examples:

• “Did you know that your face cream only works for 20 minutes after application?” (guts)
• “Did you know that TikTok checks the hash of each video in 0.3 seconds?”
• “Did you know that 80% of goods from AliExpress can be found 40% cheaper?” (product)

Why it works: “knowledge is power” works both ways: if you don’t know something, you’re weaker. The viewer stays to close the knowledge gap. The formula is universal - suitable for any vertical. Retention — 50–58%.

Formula 6: “Why doesn’t anyone talk about this?”

Mechanics: a hint of a conspiracy of silence. The information exists, but it is suppressed - and you are about to reveal the truth. The viewer feels like an insider.

Examples:

• “Why doesn’t anyone say that hyaluronic acid dries out the skin?” (guts)
• “Why doesn’t anyone say that organic reach on TikTok is free?”
• “Why doesn’t anyone say that bookmakers themselves suggest winning bets?” (betting)

Why it works: conspiracy trigger. If information is hidden, it means it is valuable. The viewer wants to be the one who “finally finds out the truth.” Works best with counterintuitive statements. Retention — 53–60%.

Formula 7: “What will happen if...?”

Mechanics: a thought experiment. The viewer presents the scenario and wants to know the result. Works better with visually demonstrated things - the video should show the process.

Examples:

• “What happens if you apply this product twice a day for a month?” (guts)
• “What happens if you upload one video to 50 accounts without uniquization?”
• “What happens if you use only free proxies for multi-accounts?”

Why it works: curiosity + anticipation of the result. The brain runs a simulation and wants to check its forecast with reality. A particularly strong hook if the result is unexpected. Retention — 51–57%.

Formula 8: “Guess how much it costs...?”

Mechanics: interactive element. The viewer mentally names the number and waits to check. Works with any numerical data - price, income, time, quantity.

Examples:

• “Guess how much this procedure costs in Korea and Russia” (gut)
• “Guess how much I spend per month on the entire stack for arbitrage”
• "Guess how many accounts one person can manage without burning out"

Why it works: game element. The spectator makes a bet and wants to check whether he guessed right. The more unexpected the answer, the higher the engagement and comments. Retention — 48–55%.

Category 3: Result Hooks - 4 formulas

The most straightforward category: show the result in the first seconds, and the viewer will stay to find out how to achieve it. It works on the basic desire to get the same thing. Particularly effective for the interior vertical (visual “before/after”) and earnings (specific income figures).

Formula 9: “I earned X for Y”

Mechanics: a specific income figure for a specific period. Specifics kill skepticism - “$3,847 in 12 days” sounds more convincing than “a lot of money in a short time.” Odd numbers are perceived as more reliable.

Examples:

• “$3,847 in 12 days on one offer - showing the whole bunch”
• "247,000 rubles per month with YouTube Shorts - without investing in advertising"
• “I saved $1,200 on creams when I found out about this product for 300 rubles” (gut)

Why it works: specific numbers trigger the “calculator” in the viewer’s head - he instantly recalculates for himself: “If it’s in 12 days, I can do it too.” Retention — 54–62%.

Formula 10: “Result in N days”

Mechanics: visual confirmation of the result with a specific timeframe. It works through proof: “this is the beginning - this is the end - this is how much time has passed.”

Examples:

• “The result after 3 days - I couldn’t believe it myself” (insides, showing skin/figure)
• “After 7 days with this bundle - here are my numbers” (office screen)
• “14 days on free traffic - that’s what happened” (income screen)

Why it works: timeframe makes the result achievable. “3 days” is not abstract, it’s concrete. The viewer thinks: “I can also try for 3 days.” Visual “before/after” enhances the effect manifold. Retention — 52–60%.

Formula 11: “Before/After - no filters”

Mechanics: visual shock from the transformation. The specification “no filters” removes the main objection – “it’s just Photoshop.” Only works with real visual contrast.

Examples:

• "My skin BEFORE and AFTER - 2 weeks, no filters, no retouching" (internal)
• “My account before and after the link - no cheating, only organic” (statistics screen)
• “My income BEFORE arbitrage and AFTER - 4 months difference” (screen)

Why it works: visual comparison is the most powerful evidence format. The brain processes images 60,000 times faster than text. “No filter” = authenticity that the audience appreciates. Retention — 56–64%.

Formula 12: “I found a way to [achieve X] for free”

Mechanics: the promise of a free solution to an expensive problem. The word “free” is one of the most powerful triggers in marketing. The viewer stays because the potential benefit is infinite at zero cost.

Examples:

• “I found a way to remove acne for free - dermatologists are silent” (guts)
• “I found a free way to get traffic without an advertising account”
• “I found a way to test creatives for free - on real traffic”

Why it works: “free” resets the entry barrier. Even if the viewer is skeptical, he thinks: “Well, I’ll listen, it’s still free.” Retention — 50–58%.

Category 4: Urgency Hooks - 4 Formulas

Urgency is a trigger that works on the fear of loss. When the viewer feels that the window of opportunity is closing, he cannot swipe in case he misses something important. Urgency hooks are especially effective for promotional videos, announcements, and time-limited offers.

Formula 13: “This only works up to...”

Mechanics: deadline. Information or opportunity will soon disappear - if you don’t find out now, then it will be too late. A specific date/event enhances the effect.

Examples:

• “This only works until the TikTok algorithm is updated - then they’ll close it down”
• “This only works until the end of March - then prices will double” (product)
• "This loophole only works until the next patch - hurry to take advantage" (gambling)

Why it works: FOMO + deadline. A potential loss is felt 2.5 times more strongly than a similar gain (Kahneman's loss theory). The viewer cannot risk missing information that will soon become unavailable. Retention — 52–59%.

Formula 14: “The platform will delete this video”

Mechanics: metahook - a link to the video itself as prohibited content. If the platform wants to delete it, it means the information is so valuable (or dangerous) that they are trying to hide it.

Examples:

• "TikTok will definitely delete this video - so watch quickly"
• “Instagram already deleted this video twice - I’m showing it for the third time”
• "This video won't last long - save it now"

Why it works: “The Streisand effect” - an attempt to hide information makes it even more desirable. The viewer feels the exclusivity of the moment: “I managed to watch it before it was deleted.” Additional bonus: encourages saving and reposting. Retention — 55–63%.

Formula 15: “Before everyone finds out”

Mechanics: exclusive information for a limited time. Few people know yet that this is an advantage. When everyone finds out, the advantage will disappear.

Examples:

• “Before everyone finds out, here’s a bundle that gives ROI 300%”
• “Until competitors find it, here is a source of traffic without a ban”
• “Before everyone finds out, this ingredient works better than Botox” (nutra)

Why it works: competitive advantage depends on the number of people who know about it. The viewer wants to be an “early adopter” and get the benefit before it becomes publicly available. Retention — 50–57%.

Formula 16: “Last chance / N days left”

Mechanics: countdown. The most straightforward trigger for urgency is a specific deadline, after which the opportunity disappears forever.

Examples:

• “3 days left - then this discount will be removed forever” (product)
• "Last chance to enter TikTok with this strategy - update in April"
• "48 hours left - then the promotional code will expire"

Why it works: a specific deadline creates a “point of no return.” The viewer is forced to make a decision: find out the details or come to terms with what he missed. Most choose the first. Retention — 48–55%.

Category 5: Storytelling Hooks - 4 Formulas

People are programmed for stories - this is an evolutionary mechanism for transferring experience. When you start telling a story, the viewer's brain switches to "listening mode" and can't stop until they know the ending. Storytelling hooks show slightly lower instant retention, but compensate for this with depth of engagement: viewers who stayed after the story hook watch the video to the end much more often.

Formula 17: “A subscriber wrote to me...”

Mechanics: social proof + the beginning of a real story. A “follower” is someone similar to the viewer, which means their story is relevant.

Examples:

• “A subscriber wrote to me: “I spent 200,000 on cream, but this helped”” (guts)
• “A subscriber wrote to me: “I lost the entire network in one night” - I’m telling you what went wrong”
• “A subscriber wrote to me: “I started from scratch and reached $5,000 in 2 months” - I’m sorting out his link.”

Why it works: someone else's story creates empathy and identification. The viewer thinks: “That could be me.” The subscriber case study format also increases the credibility of the author as an expert. Retention — 47–54%.

Formula 18: “Something happened yesterday...”

Mechanics: the beginning of a story with an unfinished plot. “Yesterday” creates a feeling of freshness and relevance. “Something” is an intrigue that requires continuation.

Examples:

• “Something happened yesterday that changed my entire approach to skin care” (guts)
• “Yesterday I discovered a bug in the TikTok algorithm - and it still works”
• "Yesterday my account got 2 million views in 6 hours - I'll tell you what I did"

Why it works: incomplete gestalt. The brain hates unfinished stories - and will hold attention until it gets a resolution. “Yesterday” adds urgency: this is not an old story, it just happened. Retention — 46–53%.

Formula 19: “I was fired / abandoned / deceived - and this is what I did”

Mechanics: a negative start with the promise of a positive turn. The rags to riches story is one of the oldest narrative archetypes.

Examples:

• “I was fired 3 months ago - today I earn more than my former boss”
• “I lost $10,000 on the first deal - and this is the best thing that happened to me”
• “The dermatologist said it was incurable - after 3 months I proved the opposite” (guts)

Why it works: empathy + anticipation of transformation. The viewer empathizes with the hero and wants to see how he overcame difficulties. Bonus: such stories generate the most comments and saves, which further boosts algorithmic reach. Retention — 48–56%.

Formula 20: “My friend / acquaintance / colleague told me a secret”

Mechanics: inside information obtained from a reliable source. “Friend” is social proof, “secret” is exclusivity.

Examples:

• “My cosmetologist friend told me what they actually use in clinics - and this is not what they sell you” (guts)
• “My arbitrage friend showed me a link that he hasn’t posted publicly for six months now.”
• “A friend from TikTok told me how the recommendation algorithm actually works”

Why it works: “first-hand information” is perceived as more reliable than any guide or article. The viewer gains access to the knowledge of the “inner circle”. The formula is especially strong when the “insider” has an authoritative position (doctor, developer, top affiliate marketer). Retention — 45–52%.

How to test hooks and scale winners

Formulas are the foundation, but without testing on real traffic they remain a theory. The same hook can show retention of 60% in gut and 35% in gambling. The only way to determine which hooks work is through A/B testing.

Hook testing methodology

A correct A/B test for hooks looks like this:

1. Prepare the body of the video - unified content after the hook (offer, CTA, visual). The body does not change between options - only the first 2-3 seconds change
2. Create 5-7 variations of with hooks from different categories. At least one shock, one interrogative, one hook result, one with urgency and one storytelling - this is how you determine which category works for your offer
3. Unique each variant through 360° Uniquizer for each test account. This is critically important: if you upload the same video (or even similar videos) to several accounts without uniqueness, the platform will link the accounts and the test will be invalid. The software will create a unique version for each account, saving the contents of the hook
4. Upload 3–5 accounts each option. More accounts = more reliable statistics, but even 3 accounts will provide representative data
5. Wait 24–48 hours and collect metrics: retention for 3 seconds, retention for 50%, views to the end, engagement (likes, comments, saves)

What to look for in the results

Key metrics for evaluating a hook:

• 3-second retention: main hook metric. If below 45% the hook is weak, change it. If above 55% the hook works, scale
• Drop-off curve: a sharp drop after 3 seconds means that the hook caught, but the body of the roller did not hold. This is not a hook problem, but a content problem
• Engagement rate: hooks that generate comments (especially arguments) receive an algorithmic boost. Provocative hooks from the “shock” and “questions” categories usually lead here
• Save rate: a high save rate means that the content is perceived as useful. Result hooks and storytelling hooks usually lead in retention

Scaling winners

Once you've identified 1-2 hooks with the best metrics, it's time to scale:

1. Create the final video with the winning hook
2. Unique via 360° Uniquizer - one version for each account in the grid. Set the number of versions equal to the number of accounts - the software will automatically create a unique file for each. Each version differs from the others at all levels of verification: pHash, audio fingerprint, metadata, neural network analysis - there will be no content links between accounts
3. Fill the entire grid with natural intervals between uploads and different publishing times
4. Monitor the metrics and at the first sign of creative burnout - retention drops below 40% - launch a new hook testing cycle

Platform differences: adaptation of hooks

One formula - different delivery on each platform. Here are the key differences:

TikTok: the “fastest” audience. The hook should be as aggressive as possible - shock and provocation work best. Text on the screen in the first seconds is mandatory - many watch without sound. The optimal hook length is 1.5–2 seconds. Fonts are large and contrasting. Trendy sounds enhance the hook.

Instagram Reels: audience is older and “smarter” - shock hooks can be irritating. Question hooks and storytelling work better. Visual quality is more important than on TikTok - blurry videos are swiped instantly. The optimal hook length is 2–3 seconds. Aesthetics and color correction are critical.

YouTube Shorts: the audience is accustomed to educational content. The result is hooks and “Did you know?” work better than provocations. Longer acceptable hook - up to 3 seconds. Less aggressive serve, more value in the first frame. Shorts less “trend-dependent” - evergreen hooks live longer.

For a large-scale flood through the UBT: create 3 versions of one video - with an adapted hook for each platform. Then uniquize each option via 360° Uniquizer for the number of accounts on each platform. One source → 3 platform adaptations → N unique versions for each grid. Automation through AI-generation and uniquization allows this cycle to be completed in hours rather than days.