What is hook rate?

Hook rate is the percentage of viewers who watched past the 3-second mark of a short-form video, divided by the total number of viewers who saw the ad. Formally: hook rate = (3-second views) / (impressions), expressed as a percentage. A video served to 10,000 people that held 2,500 of them past second three has a 25% hook rate. The metric is calculated and surfaced natively by TikTok Ads Manager, Meta Ads Manager (under “3-second video plays / impressions”), and Google Ads for YouTube Shorts.

The 3-second mark is not arbitrary. It is the empirical median decision point for swipe-away behavior across TikTok, Instagram Reels, and YouTube Shorts — the moment by which a viewer has processed the opening frame, the first audio cue, and the first text overlay, and either commits to keep watching or moves to the next video. Earlier than three seconds, the viewer hasn't finished the “is this worth my time?” calculation. Later than three seconds, the viewer is already partway invested and the drop- off rate flattens.

Hook rate is sometimes confused with thumb-stop rate, view-through rate (VTR), or completion rate. They are different. Thumb-stop is usually defined at 2 seconds and is platform-specific. VTR is a completion metric tied to the whole video, not the opening. Hook rate is specifically the 3-second-to-impression ratio, and it is the metric the cold-start algorithms care about most.

Hook rate benchmarks differ meaningfully across platforms because each feed has a different scroll velocity, a different default audio state, and a different play mechanic. The numbers below are synthesized from The Ad Bench's rubric calibration set and the publicly reported benchmark ranges from the major ad platforms.

• —TikTok.Median paid-ad hook rate sits at 25–30%. A hook rate above 40% is strong and correlates with expanded For You Page distribution. Below 20%, the FYP cold-start bucket-test fails and the algorithm caps further serves — the ad effectively dies in its first hour. TikTok's scroll velocity is the fastest of the three platforms, which compresses the decision window and pushes the kill threshold higher.
• —Instagram Reels.Median hovers around 30–40%, with 35%+ counting as strong. Reels runs hotter than TikTok because Meta's ad delivery weights creative-history signals more heavily, so an account with prior winning creatives gets a friendlier cold-start bucket. New accounts and new creatives still face the same 3-second test, just with slightly more forgiving thresholds.
• —YouTube Shorts. Median is 20–35% with 35%+ counting as strong. The loop mechanic inflates the floor because looping videos accumulate additional 3-second-plus views from the same viewer — a 12-second video that loops three times can register multiple 3-second plays per impression. That makes raw Shorts hook rate slightly more optimistic than the equivalent number on TikTok, and benchmark comparisons across the two platforms need to account for it.

Across all three platforms, the bottom-quartile videos sit below 20% hook rate and the top-decile sit above 50%. The gap between those two bands is almost entirely creative — opening line, first frame, audio cue — not targeting or bid strategy.

Hook rate is the earliest reliable distribution signal a platform can collect. By the time a video has been served to a few hundred users, the 3-second-view ratio is statistically stable enough to predict whether broader distribution will earn watch-time or waste ad inventory. Every major short-form platform uses some version of this logic to decide whether to expand a video's reach.

On TikTok, the For You Page cold-start runs a reactive bucket-test on every post: the first cohort (typically the first few hundred impressions) acts as a live experiment, and the hook rate in that first hour is the dominant input to whether the video gets pushed to a second, larger bucket. A weak hook rate in the test bucket and the video gets capped — additional spend buys diminishing reach. A strong hook rate and the algorithm expands the audience aggressively because each subsequent impression is statistically likely to convert into watch-time.

Reels and Shorts use similar but distinct cold-start mechanics. Meta's delivery system blends hook rate with account-level creative-history scores, so a brand with a track record of high hook rates gets faster early-scale even on a new asset. YouTube Shorts weights hook rate alongside loop count and like rate, which is why Shorts videos that loop well can survive a moderate hook rate that would kill the same asset on TikTok. The principle is shared across platforms: the first 3 seconds buys you the next 30 seconds of distribution.

Hook rate dies for predictable reasons. Five patterns account for the majority of below-20% openings The Ad Bench sees in the calibration set, and each one fails for a specific reason — not just a vibes- based one.

• —Brand-first openers.A logo lockup, a brand intro card, or a spoken “Hi, we're [brand] and today we're going to talk about…” signals “ad” before the viewer has any reason to care. The algorithm's ad-detection signal fires and the viewer's skip reflex follows within half a second.
• —Slow first frame.A static product shot, a slow zoom, or a fade-in on a calm background reads as “nothing is happening yet” in a feed where every other video is mid-motion. The viewer's scroll thumb is already moving by the time the frame resolves.
• —Pre-rolled lower-thirds. Caption bars, name badges, or sponsorship disclosures dropped in before the hook lands eat pixel real estate that the actual opening text needs, and they cue the viewer that this is produced content rather than feed-native creator content.
• —Low-contrast text overlay. Thin white text on a light background, or grey-on-grey caption cards, are invisible to viewers scrolling at full speed. If the text doesn't register inside the first second, the written hook may as well not exist.
• —Mismatched audio.A trending sound layered under a serious health claim, or a moody string track under an energetic product demo, creates a microsecond of cognitive friction that registers as “this feels off” — and that's enough to lose the viewer inside the 3-second window.

There are three rules that drive most of the gain from a sub-20% hook rate to a 35%+ hook rate. None of them require a bigger production budget; they require a different writing and editing discipline.

• —The first line is the viewer's problem, not your brand's solution. “Your skin looks 10 years older than it should” beats “Introducing our new vitamin C serum.” Lead with the pain. The product comes later, after the hook has earned the watch-time to deliver it.
• —First frame must contain motion plus a face or text overlay. Static frames die. A moving subject, a hand entering frame, or a kinetic text reveal in the first 200 milliseconds gives the eye something to lock onto. Sound-off viewers — more than half the audience on Reels and Facebook — are reading the screen, not listening, and they need a visible hook on frame one.
• —Audio must hit in the first 0.5 seconds with a clear beat or voice cue. Silence in the opening half-second reads as a buffering pause and triggers a skip. A musical hit, a voice line, or even a foley sound effect (a snap, a tap, a transient) on frame one anchors the sound-on viewer immediately.

The Ad Bench analyzes the first 3 seconds of every uploaded cut and scores the hook on a 0–100 scale, calibrated against the per- platform benchmarks above. The scoring rubric reads the opening line, the first-frame composition, the text overlay legibility, and the audio onset, and weights each one according to its empirical contribution to actual hook-rate performance in the calibration set.

Quick mode returns a single hook score plus a flag for any of the five common kill patterns — brand-first opener, slow first frame, pre-rolled lower-thirds, low-contrast text, mismatched audio. It runs in seconds and is designed for triage: which cuts in the batch are worth a second look, and which need a rewrite before they go anywhere near a media buy.

Deep mode does the same analysis and adds three rewritten hooks tailored to the actual product and target audience inferred from the cut. Each rewrite follows one of the six hook formulas (pattern interrupt, curiosity gap, controversy, social proof, problem- agitate-solution, value promise) and is scored against the same benchmark so you can compare predicted hook rate before vs. after the rewrite. The goal is to make the gap between a 15% hook rate and a 40% hook rate solvable in the writing room, not in the edit bay after the campaign has already underperformed.