Hook Rate in 2026: The 3-Second Metric That Decides Meta Ads

Hook rate is the Meta video metric that decides whether your ad ever gets a real audition. It's the percentage of impressions that turn into 3-second video views, and Meta's Andromeda ranking system now leans on it hard when picking which creatives to scale. Most accounts we audit have never set it as a custom column, so they end up scaling video that loses 80% of viewers in the first second. This guide covers the formula, 2026 benchmarks, the diagnostic flow that separates a thumb-stop problem from a payoff problem, and the 11 levers that actually move the number.

TL;DR: Hook rate = (3-second video views ÷ impressions) × 100. Aim for 25% as a baseline, 30%+ as good, and 35%+ as scalable. Pair it with a 40-50% hold rate. Below 25%, rework the first frame. Above 35% with weak hold, rewrite the payoff. Andromeda rewards creatives that earn attention in the first 3 seconds — set the column, watch it daily, treat it as an upstream KPI.

What hook rate actually measures

Hook rate is the share of impressions that become 3-second video views. The formula is fixed:

Hook rate = (3-second video views ÷ impressions) × 100

Worked example. An ad set serves 100,000 impressions and Meta records 28,000 3-second views. Hook rate is 28%. That's the universe. Anyone who didn't make it to second three never saw your offer, demo, testimonial, or CTA. You're optimizing for that 28% slice for everything downstream, including CTR, conversion rate, and ROAS.

Meta counts a 3-second view in the video metrics documentation when the video plays for at least three continuous seconds, including in autoplay. Impressions are counted at first render.

This is different from thumb-stop ratio, which uses a 2-second threshold and is an internal diagnostic rather than a column. It's also different from video watch time, which measures average duration. The 3-second metric is a binary check, a yes/no on whether you earned the audition.

One subtlety: the denominator is impressions, not reach. If frequency creeps and one viewer sees the ad five times, all five count. That's why the number sags as creatives age. We cover that decay curve in our ad fatigue guide.

Why Meta's Andromeda weights it heavily in 2026

Andromeda is Meta's machine-learning ranking system, rolled out in late 2024 and now the default for ad delivery. The Meta engineering Andromeda announcement describes a 10,000x increase in model complexity, with sharper signals about which creatives keep users engaged in-feed.

In practice: early engagement signals, including the 3-second view, matter more for distribution. If your creative loses 75% of impressions before second three, Andromeda reads that as low engagement quality and throttles delivery. If it holds 35%+, the system gives the creative more impressions in similar contexts. The ranking runs continuously, ad set by ad set, placement by placement.

Stack this with Advantage+ Creative and the picture sharpens. Advantage+ tests micro-variants of the same asset, and the variants that win hold viewers past the 3-second wall. Without the upstream filter, the rest of the optimization stack is sorting noise.

This is why hook rate stopped being a vanity number and became a leading indicator. When we look across thousands of high-spend in-market ads on adlibrary saved-ads collections, creatives still serving past day 21 almost universally hit 30%+ at launch. The ones that died at day 7 launched at 15-20%.

2026 benchmarks: what counts as good

There's no single industry number, since Meta doesn't publish one, but practitioner data has converged. Below are the working bands we use, cross-referenced with public benchmarks.

Sources: the Vaizle Meta benchmarks report puts the median around 25% across mixed verticals. Triple Whale's creative report treats 30% as the scaling threshold for DTC. Sovran's analysis tracks the 35% line as the cut-point for top-quintile creatives. Billo's UGC benchmarks report 25-30% as typical for paid UGC.

Vertical adjustments matter. B2B and considered-purchase categories run lower (15-25% can be acceptable when the hold rate is exceptional). DTC apparel and impulse categories should clear 30% before you scale. Reels placements typically beat Feed by 5-8 points but suffer worse hold rate, since swipe-friction is lower in both directions.

For a sanity check on whether your 25% is good for your category, run the comparison on adlibrary's ad timeline analysis. Competitors still running the same creative for 60+ days are clearing the bar Meta cares about.

How to set up hook rate as a custom metric in Ads Manager

Meta still doesn't ship hook rate as a default column. You have to build it. The Meta custom metrics documentation covers the mechanics, but here's the path most teams want.

In Ads Manager, open any campaign view. Click Columns → Customize Columns. Scroll to Custom Metrics → Create Custom Metric. Name it "Hook Rate". Set the formula:

(Video Plays at 3s ÷ Impressions) × 100

Format as percentage with one decimal. Drag it into your default preset between CTR and CPM.

Two pairings compound the value. Add Hold Rate: (Video Plays at 75% ÷ Video Plays at 3s) × 100. The first tells you who stopped scrolling, the second who stayed for the payoff. Together they isolate where attention dies.

Add ThruPlay rate as a third: (ThruPlays ÷ Impressions) × 100. ThruPlay is Meta's billable view event for the video views objective, at 15 seconds for ads under 15s, or full completion for shorter assets. Useful as a tiebreaker.

For account managers handling multiple clients, save the preset and propagate it through agency workflows and campaign benchmarking routines.

Hook rate vs hold rate vs ThruPlay: a comparison

These three video metrics get conflated. They measure different things and answer different questions.

The decision tree is simple. Look at hook rate first. If it's below 25%, fix the opener. Nothing else matters because the audience never arrived. If hook rate clears 25% but hold rate is below 40%, the opener is bait-and-switch. If both clear (hook 30%+, hold 45%+) but CTR is weak, your offer or CTA is the problem, not the creative.

Hook rate alone is a noisy signal. Hook × hold is the actual creative quality score. We've seen accounts with 38% hook and 28% hold scale fast and crash inside 14 days. The reverse pattern (22% hook, 55% hold) is more fixable: a different first frame is all the asset needs.

For agencies running multiple accounts, building hook/hold/CTR triplet dashboards inside creative testing workflows cuts diagnosis time from hours to minutes.

Step 0: find the hook patterns from in-market ads first

Before briefing a single new creative, audit what's already winning in your category. This is the step most teams skip and then wonder why their hook rate sits at 18%.

The cheap path: open adlibrary saved-ads and pull the top 50 ads in your vertical running 30+ days. Anything still serving at 30 days has cleared Meta's quality bar, which almost always means above 25% hook rate. Filter the saved set by media type for video only. Then use AI ad enrichment to extract the hook tag from each ad: question hook, demo hook, character POV, transformation reveal, pattern break.

You're looking for the dominant archetype. If 30 of the 50 top ads use a "messy reality" cold open (UGC creator at home, no makeup, talking to camera), that pattern is winning impressions right now. The algorithm rewards it because viewers reward it. Your brief should ride that wave, not invent a new one.

Cross-reference with ad timeline analysis. Ads live across multiple iterations (V1, V2, V3 of the same hook) are the ones the brand has confirmed work, because they're spending production budget to remix them. Higher-confidence signal than a 30-day check alone.

For deeper category mapping, creative strategist workflows and ecommerce product research lay out the swipe-to-brief loop. Step 0 is not about copying, but entering the conversation Meta's already optimizing inside.

11 ways to boost hook rate

These are the levers that move the number, ranked by how often they work in audits we run. Test in order.

Frame 1 first. The default fail mode is opening on a logo card, an empty room, or a wide shot. None of those tell the viewer what's in it for them in 0.3 seconds. Start on the most interesting moment, what editors would put as the third or fourth scene. Drag it to position one. Teams find the metric jumps 5-8 points from this single move.

Audio-off design matters more than people think. Meta's feed analytics show 80%+ of video impressions play with audio off in Feed and Stories. If your hook depends on a voiceover, you're optimizing for the wrong audience. On-screen text in the first 0.5 seconds is non-negotiable for cold traffic. The text should restate the hook claim, not subtitle the dialogue.

Pattern interrupt, like a hard cut or graphic glitch, works because the feed is a sea of slow pans and steady talking heads. Anything different earns the half-second that gets you to the 3-second milestone.

For UGC ads, the highest-impact move is character POV: a creator's face filling the frame, eye contact, immediate spoken claim. This pattern dominates the top 10% of results in our ad creative testing data, especially for DTC. The algorithm reads "person looking at camera" as social content, and the instinct to scan faces is older than feeds.

For e-commerce, video ads for ecommerce stores walks through a shot list that puts the product into frame inside 1.5 seconds without sacrificing the hook claim.

Diagnostic flow: low hook + high hold vs high hook + low hold

The two-by-two of hook rate × hold rate isolates which part of the creative is broken. Most teams treat "low video performance" as a single problem and rebuild the whole asset. Wrong move. Three out of four times only one half of the creative is failing.

Quadrant 1: Low hook, low hold (under 20% / under 30%). The whole creative is misaligned with the audience. Don't fix it. Replace it. This usually means wrong angle, wrong format, or wrong audience entirely. Go back to Step 0 and rebuild from category swipe.

Quadrant 2: High hook, low hold (30%+ / under 35%). Bait-and-switch. The opener over-promises, the body under-delivers. Rewrite the middle 5-15 seconds without touching the opener. Pull the strongest payoff moment forward, cut filler, restructure the demo, sharpen the CTA. Clickbait-y openers ("you won't believe...") that don't connect to the actual product also produce this pattern.

Quadrant 3: Low hook, high hold (under 25% / 50%+). Thumb-stop problem. People who watch love it, but you can't get them to start. This is the highest-impact quadrant. Rebuild only the first 1.5 seconds. New first frame, new opening claim, same body. The metric often jumps 10-15 points from a 4-hour edit.

Quadrant 4: High hook, high hold (30%+ / 50%+). Scale. Push spend, expand to similar audiences, build Advantage+ Creative variants, sequence into a retargeting flow.

This diagnostic also exposes a common mistake: killing winners too early. A creative with 28% hook and 55% hold is a workhorse. Hold rate that high means the learning phase hasn't finished and the audience that watches is a strong fit. Use the learning phase calculator to confirm you're not killing pre-resolution. For broader diagnosis, how to analyze ad performance walks the funnel.

Common mistakes that tank hook rate

A short list of patterns we see weekly. Each is fixable in under an hour.

Opening on a logo or brand card. This burns the first 1-2 seconds on something the viewer doesn't care about yet. Brand recall happens on click and landing page, not in frame 1. Move logos to a corner watermark or last frame.

Voiceover-dependent hooks with no text overlay. Audio off, hook gone. If the viewer needs to hear "Tired of slow shipping?" to understand the ad, 80% never get it. Always show the claim in text by 0.5 seconds.

Wide establishing shots. Cinematic openers don't work in feed. The viewer is scrolling through a 6-inch screen at thumb speed. A wide shot reads as background noise. Open tight on the subject.

Generic "stock-feel" first frames. Pristine white-background product, a model on a beach, a slow zoom on a smiling face. Algorithms have learned these mean "ad" and viewers have learned to scroll past them. Use the messiest, most native-looking frame you have.

Burying the differentiator past second 5. If your USP is in second 8, you're showing it to fewer than half the people who started the video.

Recycling assets across formats. A 1:1 Feed video pasted into 9:16 Reels has letterboxing. The black bar reads as low-quality and crushes the metric by 5-10 points. Re-cut for the placement. See meta ad sizes 2026 for native ratios.

Not testing 3+ openers per concept. One opener equals one data point. The same body cut three different ways often produces results ranging from 18% to 38%. The opener is the highest-variance element of the creative. Test it, don't guess. We've codified this in our automated split testing approach.

Letting creative fatigue run unchecked. The metric decays 5-15% per week of unchanged spend on the same audience. Rotation, not duration. Ad fatigue covers refresh cadence.

Frequently asked questions

What is a good hook rate on Meta?

A good hook rate on Meta in 2026 is 30% or higher. 25% is the working baseline, 30%+ qualifies as good, and 35%+ is the threshold most performance teams use before scaling spend. Below 25% the opener needs rework. The exact target shifts by vertical: DTC impulse categories should clear 30%, while B2B and considered-purchase categories can run sustainably at 20-25% if hold rate is strong.

What is the difference between hook rate and CTR?

Hook rate measures attention (3-second video views ÷ impressions), while CTR measures action (link clicks ÷ impressions). One tells you whether the creative stopped the scroll. The other tells you whether the offer earned a click. A creative can hit 35% hook rate and 0.4% CTR (strong attention, weak conversion path), meaning the opener works but the offer or CTA doesn't. They're sequential filters, not interchangeable. Use the CTR calculator for clean ratio math.

How do I calculate hook rate?

Divide 3-second video views by impressions and multiply by 100. Formula: hook rate = (3-second video views ÷ impressions) × 100. Set it up as a custom metric in Meta Ads Manager via Columns → Customize Columns → Create Custom Metric. Name it "Hook Rate", use video plays at 3s ÷ impressions, format as percentage. Save the column preset to your default view.

What is a bad hook rate?

A bad hook rate on Meta is anything below 20%. At that level, more than four out of five impressions never become a viewer. Below 15%, the ad is broken: Andromeda is throttling distribution, costs are inflating, and learning phase will not resolve. Rebuild the first frame before changing audience or budget.

What is a good hook rate for Reels?

A good hook rate on Reels is 30-40%, typically 5-8 points higher than Feed for the same creative. Reels' swipe-friction is lower in both directions: viewers scroll past faster, but they also pause faster on anything that breaks the rhythm. Vertical 9:16 native composition, immediate motion, and on-screen text in the first 0.5 seconds are the dominant levers. Above 45% is excellent but watch hold rate. Reels viewers swipe out faster too.

Bottom line

Hook rate is the upstream filter for every Meta video metric. Set the column, treat 25% as the floor and 30%+ as the bar to scale. Pair with hold rate to isolate thumb-stop vs payoff problems. The fastest lift comes from rewriting frame 1, not the whole asset.