Hooks for Shorts, Reels and TikTok: Choose, Write, Test
Choose, write and test a hook for YouTube Shorts, Reels and TikTok, using each platform's own retention metrics and five complete shot plans.
On this page 9 sections
A hook is everything a viewer sees, reads and hears in the opening seconds of a Short, Reel or TikTok, and its job is to make them stay instead of swiping. The platforms describe that window as "the first few seconds" (YouTube), "the first 3 seconds" (Instagram) and "ideally within the first 5 seconds" (TikTok). To get a hook that holds, choose it from what your video actually delivers, put that payoff or the viewer's problem into the first frame, the on-screen text and the first spoken line, then check it against each platform's own measure of who stayed. No hook guarantees views, and the platforms' own help pages give no target score, so your own recent posts are the benchmark.
Need raw material first? The viral hook library has 52 hook templates, and the viral hook generator drafts opening lines for a topic you enter. This guide covers what a list can't: choosing between hooks, turning one into a shot plan, and checking whether it worked.
What YouTube, TikTok and Instagram say about the first seconds
| Platform | What it tells creators | The "did they stay?" metric | Where to find it |
|---|---|---|---|
| YouTube Shorts | "Capture attention in the first few seconds to prevent viewers from scrolling" (YouTube blog, April 2025) | Stayed to watch: "The percentage of times viewers stayed to watch past the initial seconds of a Short" | YouTube Studio → Analytics → Content → Shorts |
| TikTok | "Consider having an attractive element early on in your video that sparks interest, ideally within the first 5 seconds" (Creator Academy) | Retention rate, which TikTok says helps you "find drop-off points", plus average watch time and watched full video | TikTok Studio → Analytics → tap a post |
| Instagram Reels | "Make sure the first 3 seconds of your reel are engaging, so that people don't move on" (Instagram for Creators) | Skip rate where your insights show it, which Meta's developer documentation defines as "the percentage of views from people who skipped during the first 3 seconds"; otherwise average watch time | Profile → Reels tab → the reel → View insights |
Definitions checked October 19, 2026; the platforms rename these metrics from time to time. YouTube's help pages also call the Shorts measure "How many chose to view", described as "the percentage of times that viewers viewed your Shorts versus swiped away", so look for either label.
Two details change how you should read these numbers:
- Views don't measure the hook. Since August 24, 2026, YouTube counts a view "the moment a video starts to play", Shorts included, while an engaged view counts only when a viewer stays "past the initial seconds" (YouTube Help). Instagram likewise counts a reel view each time it "starts to play or replay". A Short that most people swipe past can still pile up views.
- There is no official target. YouTube describes an engaged view only as one that keeps playing for "some amount of seconds" after the first frame, and none of these help pages gives a good or bad percentage. Benchmarks quoted elsewhere come from third-party samples, not from the platforms. Compare each post with your own recent posts of the same format and length. YouTube's analytics tips make a similar point: "it's key to still compare videos of the same type."
Looking for a hook video for Shorts?
If you searched for a "hook video for Shorts" because you want a ready-made clip to paste in front of your video, there is a catch. The hook is not a separate clip. It is the opening of your own video, and it has to promise what the rest delivers. A borrowed viral clip causes three problems. You may not have the rights to it. Instagram lists "unoriginal content that is largely repurposed from another source with only minor, immaterial edits" among content it may not recommend (Instagram Help). And if viewers stayed for the clip rather than your topic, your retention graph will show them leaving where the clip ends. Open with your own subject instead: the finished result, the problem, or the moment your story turns, as in the shot plans below.
How do you choose a hook for a Short or Reel?
Start from the payoff, not from a list. Write one sentence that says what a viewer has by the end of your video, then let the type of payoff pick the pattern:
| If your video delivers... | Use this pattern | The first frame shows |
|---|---|---|
| A visible result: a finished photo, dish, repair or makeover | Result first | The finished result |
| A fix for an everyday annoyance | Problem on screen | The problem happening |
| The answer to something people wonder about | Question answered fast | The thing being asked about |
| A correction to a common belief | Claim plus source | The mistake, then the proof |
| A story with an ending worth waiting for | Turning point first | The moment the story turns |
If two patterns fit, draft both and keep the one whose promise the video pays off most completely. If none fits, the problem is the video, not the hook: a hook can only promise what the video already contains.
Five hook examples for Shorts and Reels, with complete shot plans
Each plan gives the picture, the on-screen text and the voice for the opening and the payoff. Timings are examples; adjust them to your footage. None of the five needs a face on camera, and every shot should use footage you filmed, licensed stock or visuals you generated.
1. Result first: phone product photos for a small shop
| Time | Picture | On-screen text | Voice or sound |
|---|---|---|---|
| 0:00-0:01 | The finished photo of a ceramic mug fills the screen | "Shot on a phone, by a window" | "This is a phone photo, taken next to a window." |
| 0:01-0:03 | Pull back: the mug on white paper beside the window, the phone propped on books | "The whole setup" | "Here's the whole setup." |
| 0:03-0:15 | Three steps, one shot each: move the mug so the window is beside it, not behind it; stand a white card on the shadow side; tap the mug on screen to focus | "1. Window to the side 2. White card 3. Tap to focus" | One short line per step |
| 0:15-0:20 | Split screen: the first attempt, with the window behind the mug, next to the finished photo | "Before / after" | "Same phone. Same window." |
| End | Cut back to the finished photo, so the loop restarts on the result | None | None |
The first frame is the promise and the rest is the method. It fails when the result is no better than what viewers already get, because then the first frame promises nothing new.
2. Problem on screen: the charger cable that falls behind the desk
| Time | Picture | On-screen text | Voice or sound |
|---|---|---|---|
| 0:00-0:01 | Close-up: a charging cable slides off the desk edge and drops out of sight | "Charger fell behind the desk again?" | The sound of the cable hitting the floor |
| 0:01-0:03 | A binder clip snaps onto the desk edge | "One binder clip" | "Clip a binder clip to the edge of the desk." |
| 0:03-0:10 | The cable goes through the clip's handles; the phone is unplugged; the cable stays on the desk | "Stays put" | "Thread the cable through the handles. Unplug, and it stays." |
| 0:10-0:15 | Wide shot: three cables in three clips | None | "One clip per cable." |
Viewers who have this problem recognize it before a word is spoken, which also makes the opening work with the sound off. It fails when the first frame shows a problem your audience doesn't have.
3. Question answered fast: why flamingos are pink
A faceless explainer, built from stock or generated visuals and a voiceover.
| Time | Picture | On-screen text | Voice or sound |
|---|---|---|---|
| 0:00-0:02 | An adult pink flamingo, then a white-gray chick | "Flamingos aren't born pink" | "Flamingo chicks hatch white-gray. So where does the pink come from?" |
| 0:02-0:05 | Flamingos feeding in shallow water | "Their food" | "Their food." |
| 0:05-0:14 | Carrots and tomatoes, then algae and brine shrimp | "Carotenoids: the pigments in carrots" | "Pigments called carotenoids, the kind that make carrots orange, are in the algae that brine shrimp eat. A flamingo's body turns them into pink feathers." |
| 0:14-0:20 | A pale young flamingo beside an adult | "Source: Smithsonian's National Zoo" | "Young flamingos take several years to turn fully pink." |
The facts in this plan come from the Smithsonian's National Zoo. The question is answered within seconds, as promised, and the reason to keep watching is the detail that follows. It fails when the answer is held back until the last second with filler in between.
4. Claim plus source: don't rinse raw chicken
| Time | Picture | On-screen text | Voice or sound |
|---|---|---|---|
| 0:00-0:01 | Raw chicken held under a running tap; the frame freezes | "Stop. Don't rinse raw chicken." | "Stop. Rinsing raw chicken doesn't make it safer." |
| 0:01-0:05 | The USDA page on screen, one sentence highlighted | "USDA: washing raw poultry is not recommended" | "The USDA says washing raw poultry can spread bacteria from its juices to other food, utensils and surfaces." |
| 0:05-0:12 | Chicken goes from the pack straight into the pan; hands and board washed | "Pack to pan. Then wash hands and board." | "Rinsing doesn't destroy bacteria. Cooking does." |
| 0:12-0:18 | A thermometer in the thickest part of the breast reads 165°F | "Cook to 165°F" | "Cook all poultry to 165 degrees, checked with a food thermometer." |
The claim and the cooking temperature come from the USDA's Chicken from Farm to Table guidance. A contrary claim gives people a reason to stop; the source on screen gives them a reason to believe it. If you can't show a source, use another pattern.
5. Turning point first: a short horror story
A faceless story Short. It is fiction, so say so in the caption.
| Time | Picture | On-screen text | Voice or sound |
|---|---|---|---|
| 0:00-0:02 | A handwritten note on a fridge, lit by a phone screen | "Rule 3: never open the door after 10 pm" | "The babysitting note had three rules. At 10:04, someone knocked." |
| 0:02-0:05 | A clock reads 10:04; the front door in shadow | Captions | Three knocks |
| 0:05-0:40 | Back to that afternoon: the job, rules 1 and 2 | Captions | Narration |
| 0:40-0:55 | Back at the door: the reveal | Captions | Narration |
| End | The note again, with a line that wasn't there before | None | Silence |
TikTok's Creator Academy suggests "structuring your videos like a story with a beginning, a middle, and an end". Opening at the turning point gives the viewer the question first and the beginning second. It fails when the story never reaches the scene it opened with.
How to write the hook: picture, text and voice must agree
- Picture. Start on the subject, not on a logo or a greeting. TikTok calls a strong opening shot "vital to grabbing a viewer's attention" and advises "having your image centered and not having it covered by any buttons on the screen".
- On-screen text. State the promise in a few words that can be read before the next cut. According to TikTok's Creator Academy, over 30% of TikTok survey respondents watch videos on mute and rely on captions for context (the article does not date the survey), so the text has to make sense without the voice.
- First spoken line. Say the same promise in plain words, starting on the topic. "Hey guys" and "in this video" use up the seconds all three platforms point to.
- Title and caption. These label the video for search, your profile and the post description. They are not the hook. In the Shorts feed a Short starts playing by itself, and YouTube counts a view "the moment it starts to play", so the opening has to work before anyone reads the title.
Check the hook before you publish
- Watch it muted. Do the first frame and the text say what the video is about?
- Pause on the first frame. Is the subject clear, and is anything important hidden under the platform's buttons or caption area?
- Find the payoff timestamp. Write the promise in one sentence and note the second where the video delivers it. If there is no such second, rewrite the opening or the video.
- Check every claim in the opening. A number, "scientists say" or "after testing" stays only if you can show the source or the test.
- Cut the warm-up. Delete everything before the first useful frame.
- Use YouTube's Get feedback if you see it. YouTube's Help Center describes a Get feedback button in the Shorts upload flow that gives "an analysis of your video's hook, pacing, and more" (YouTube Help). Treat it as a second opinion, not a verdict.
- Export a clean file for each platform. Instagram's creator FAQ says a third-party watermark on your reel "may be impacting its reach", so don't post a TikTok download, watermark and all, as a Reel.
How to test hooks with your own analytics
You can't split-test a Short's opening inside YouTube. Its A/B testing page says plainly: "You are not able to A/B test Shorts, Scheduled Lives, and Premiere videos at this time" (YouTube Help). A hook test therefore means comparing separate posts, with care:
- Set a baseline. Note the hook metric (stayed to watch, skip rate, or the start of TikTok's retention rate graph) for your last 10 posts, or as many as you have, of the same format and similar length on that platform. Length matters: TikTok notes that watched full video "can also fluctuate depending on your video's total length".
- Change one thing. Keep the topic, length and payoff. Change only the opening, for example result first versus problem on screen.
- Post a few variants, spaced out. YouTube's spam policy says "posting a few variations of a video is ok" but bans using automated tools to flood YouTube with repetitive content (YouTube Help). On Instagram, a trial reel is "shown to non-followers first", and its metrics appear about 24 hours after you share it (Instagram for Creators).
- Read the numbers in order. First the hook metric (did they stay?), then the payoff metric (average percentage viewed on YouTube, average watch time or watched full video on TikTok, average watch time on Instagram), then the result you care about, such as subscribers, follows, profile visits or enquiries. YouTube's own performance FAQ says "relative watch time is more important for short videos".
- Repeat before you decide. Organic posts are not a controlled experiment: audience, timing and distribution differ from one upload to the next. Adopt a pattern once it beats your baseline on several videos, not one.
Reading a test (an example with round numbers, not real data). Two Shorts of the same length on the same topic:
| Metric | Short A: result first | Short B: "You won't believe this" |
|---|---|---|
| Stayed to watch | 60% | 75% |
| Average percentage viewed | 50% | 30% |
| Per 100 starts: viewers who stayed × share watched | 60 × 0.50 = 30 | 75 × 0.30 = 22.5 |
B won the first seconds and lost the rest. Per 100 starts it produced less total watching (a rough comparison that ignores the seconds watched by people who swiped). When your numbers look like B's, check whether the opening promised something the video doesn't deliver. The reverse pattern, a low stayed-to-watch with a high average percentage viewed, says the people who stay like the video and the opening is what needs work.
Why viral hooks fail
- Clickbait that wins the swipe and loses the viewer. YouTube tells creators to avoid titles and thumbnails that are "Deceiving, misleading, clickbaity or sensational" (YouTube Help). Instagram's creator FAQ asks for "authentic (ie. not clickbait) engagement" and lists clickbait among the content users say they dislike. In your analytics, it looks like Short B above.
- A hook the video doesn't pay off. The opening promises three mistakes and shows two, or the answer arrives at second 55 after filler. At its worst this is what YouTube's spam policy calls "malicious clickbait": misleading titles, thumbnails, descriptions or imagery for a video "that does not deliver what was promised". An accidental broken promise is easier to fix: look for the drop in the retention graph. TikTok's Creator Academy suggests checking "what content you covered during the drop-off point".
- A title that is not a hook. The clever line sits in the title or caption while the video opens on a logo, a greeting or a slow pan. YouTube's title advice still applies ("Be accurate", and "Viewers may only see part of your title"), but in the feed the opening frames do the work. Write titles for search and the opening for the stay.
- Borrowed credentials. Templates written as "after testing 10 tools" or "scientists just found" carry claims you must be able to back. Rewrite the opener if you can't.
- A video that is all hook. YouTube's blog notes that hooking viewers early is "great for pulling in views, but views don't automatically equal fans", and TikTok's 2020 explainer of its For You feed calls finishing a longer video "from beginning to end" a strong indicator of interest. The hook earns the first seconds; the rest of the video has to earn the rest.
From shot plan to finished video
Write the shot plan before you film or generate anything, then build the video in that order: the planned first frame first, the payoff where the plan puts it. If you make faceless videos, FluxNote turns a topic or script into a finished video with an AI-written script, voiceover, captions and AI or stock visuals, and its full video editor allows unlimited re-renders, so you can rework the opening and render again before you post. It generates new videos; it does not cut hooks out of existing long videos or podcasts. Write the title separately with the YouTube title generator, and the post description with the TikTok caption generator, which writes descriptions, not timed subtitles. Then run the checks above before you publish.