Hook rate, explained.
By Denzel Geng · Updated November 1, 2026
Hook rate is the percentage of impressions that watch the first 3 seconds of your video ad. Formula: 3-second video plays ÷ impressions × 100. A good hook rate on Meta is 25–35%: the largest published dataset (Billo, 88,329 sales-objective video ads, H1 2026) averages 25.44%, a 479-ad 2026 sample of US accounts (Biddyco) has a median of 20.9%, 30% and above is strong, and under 20% means the opener needs replacing.
Hook rate isn't a Meta metric: it's a custom metric you build in Ads Manager from Meta's 3-second video plays ÷ impressions. Some teams use a 2-second “thumb-stop” view instead — pick one and stay consistent. The bands above are the working ranges Meta buyers use; Meta itself publishes no target.
Hook rate = 3-second video plays ÷ impressions × 100.
The hook rate formula (and hold rate).
Hook rate needs two numbers from Ads Manager: 3-second video plays and impressions. Hold rate adds a third, ThruPlays, which Meta counts when a video plays to completion or for at least 15 seconds.
Hook rate
3-second video plays ÷ impressions × 100
Of everyone served the ad, the share who watched past the first 3 seconds.
Hold rate (funnel)
ThruPlays ÷ 3-second video plays × 100
Of the people the hook stopped, the share who stayed to 15 seconds or the end. Isolates the body of the video from the hook.
Hold rate (per impression)
ThruPlays ÷ impressions × 100
Of everyone served, the share who reached 15 seconds or the end. Mixes hook and body together, so it reads much lower.
Worked example
- Impressions
- 40,000
- 3-second video plays
- 10,000
- ThruPlays
- 1,500
- Hook rate
- 10,000 ÷ 40,000 × 100 = 25%
- Hold rate (funnel)
- 1,500 ÷ 10,000 × 100 = 15%
- Hold rate (per impression)
- 1,500 ÷ 40,000 × 100 = 3.75%
The two hold rates differ by exactly the hook rate: 25% × 15% = 3.75%. That's why a quoted hold rate means nothing until you know which denominator it uses. Biddyco's 4.6% median hold rate, for example, is the per-impression version; a funnel hold rate for the same ads would be several times higher.
How to add hook rate as a custom metric in Ads Manager.
Hook rate isn't a native Meta column. Meta's list of video ad metrics has 3-second video plays, ThruPlays and video plays at 25–100%, but no hook rate, so Ads Manager won't show it until you build it as a custom metric. It takes two minutes, once:
Open Columns → Customize columns
In Ads Manager, click Columns above your ads table and choose Customize columns from the dropdown.
Click Create custom metric
This opens the custom metric creation form.
Build the formula
Pick 3-second video plays from the Select metric dropdown (or start typing it), add the division operator, then pick Impressions. Meta needs at least two metrics and one symbol in a formula.
Name it and set the format
Name it “Hook rate” (hyphens, underscores and spaces are allowed), add an optional description, and choose Percentage (%) as the format.
Choose who can see it, then Create metric
“Only you” keeps it editable; “Everyone with access to this ad account” shares it, but then the formula can't be edited or deleted. Click Create metric, or Create and add another to build hold rate straight after (ThruPlays ÷ 3-second video plays).
Add the column and sort
Add the new column to your view and sort your video ads by it. The spread between your best and worst hooks is usually the most useful number on the screen.
Meta's field names to use
- 3-second video plays
- Plays of at least 3 seconds (or about 97% of a video shorter than 3 seconds), counted once per impression; replays are excluded.
- Impressions
- The number of times the ad was on screen. The denominator for hook rate.
- ThruPlays
- Plays to completion, or for at least 15 seconds.
- Video plays at 25%, 50%, 75%, 95%, 100%
- Plays that reached that share of the video's length, including plays that skipped to that point. Use them to see where the body loses people.
- 2-Second continuous video play
- Plays of 2 continuous seconds or more. Build a 2-second hook rate from it if you need to compare with TikTok's 2-second convention.
Field names and steps checked against the Meta Business Help Center on November 1, 2026.
On TikTok the convention is a 2-second “thumb-stop” view instead of Meta's 3-second play, so a TikTok hook rate will read higher for the same creative — never compare the two platforms' numbers directly.
Hook rate benchmarks: what's a good hook rate?
A good hook rate for a Meta video ad is 25–35%. Meta publishes no target, so that band comes from the three numbers that can be checked: the largest dataset we could verify, Billo's H1 2026 benchmark of 88,329 sales-objective Meta video ads, averages 25.44% (cold and warm traffic mixed); Biddyco's September 2026 cut of 479 Meta video ads across 17 US accounts ($10.1M of spend, June–August 2026, each ad over $1,000) puts the median at 20.9%, the middle half between 14.2% and 30.4%, and the top 10% at 44.8% or better; and the one independent study that names a target, Funnel Insiders' 403-ad analysis, calls 38% strong. The working bands Meta buyers use fall between those points:
Under 20%
Weak. The first frame or first line isn't stopping the scroll; most of the audience never sees the rest.
Replace the hook outright; nothing downstream can compensate for a hook nobody sees past.
20–25%
Below the cross-industry average (25.44%).
Test new opening frames and first lines against the current one — change only the first 3 seconds.
25–35%
Good. Around or above the average, and the range most scaling accounts sit in.
Keep iterating hook variants, and check hold rate before touching the rest of the video.
30–40%
Strong. The hook is doing its job.
Shift attention to hold rate and CTR; the hook is no longer your bottleneck.
Over 40%
Elite. Usually retargeting, a native-format disguise, or a very recognisable brand.
Scale it, and reuse the hook pattern on your next creative.
Hook rate benchmarks by source
These are reference values each company drew from its own clients' or customers' ads. They are not Meta metrics, and Meta publishes no hook-rate target. All use Meta's 3-second basis unless the row says otherwise.
Billo ↗Aug 6, 2026
25.44% average; categories from 21.32% to 28.79%
Meta video ads, sales objective; 88,329 ads, 14 industries, Jan–Jun 2026
Biddyco ↗Sep 27, 2026
20.9% median; middle half 14.2–30.4%; top 10% at 44.8%+
Meta video ads; 479 ads, 17 US accounts, $10.1M spend
Hawky ↗Mar 24, 2026
Under 15% a problem; 15–24% below average; 25–30% baseline; 30–39% good; 40%+ elite
Meta (Facebook and Instagram) video ads; cites aggregated Vaizle and MHI Growth Engine data
Hawky ↗Mar 24, 2026
Under 20% too slow; 20–29% below average; 30–35% solid; 35–40% top quartile; 40%+ top 10%
TikTok, on a 2-second view basis
Exposure ↗Jun 16, 2026
Under 20% weak; 20–25% below par; 25–35% good; 30%+ strong; 40%+ elite
Meta video ads; the agency's own launches, labelled directional
AdMakeAI ↗Jun 23, 2026
20–25% workable; about 25% average; 30%+ strong; 40%+ elite
Meta; a synthesis of other published benchmarks
Motion ↗Undated (read Nov 1, 2026)
20–40% generally solid; varies by industry and audience
Meta video ads (glossary definition)
Funnel Insiders ↗Apr 3, 2024
About 38% (0.376) as a target; no correlation with CPA or ROAS found
Meta; 403 ads, 26 accounts of one info-product business, Q1 2024
Our read: start with your own account's median, then look at the ranges. Sort your last 90 days of video ads by hook rate and take the middle one; that's the bar a new hook has to beat. The guides above put “good” between 25% and 30% and call 40%+ elite, while the two datasets that disclose their sample put the typical ad lower, at a 20.9% median and a 25.44% average.
Why published benchmarks disagree
Every blog quotes different numbers because four things move the metric before creative quality does:
The counting window
3-second plays vs 2-second thumb-stops — the shorter window always reads higher.
Placement mix
Reels, Stories and Feed can read several points apart for the same ad — in either direction — so a shift in placement mix moves the blended number.
Audience temperature
Retargeting audiences recognise you and stop; cold prospecting traffic doesn't.
Format and vertical
Categories differ (see the industry table below), and operators commonly report that UGC and native-style openers outpace polished brand spots.
Treat the bands as a first read, not a pass mark: Billo's 25.44% mixes cold and warm traffic with no placement split, and a cold prospecting ad on Reels can sit several points under a retargeting ad in Feed with the same opener. See the industry breakdown below for where your category sits. Then the comparison that matters most is your own hooks against each other — test variants on the same video and keep the highest.
How to use hook rate when making ads.
Use hook rate to decide one thing: whether to keep the first 3 seconds or replace them. Read it per placement and per audience, against your own best hook, then act on the band:
Under 20%: replace the first 3 seconds
Keep the body, the offer and the ending; swap only the opening frame and the first line. Below 20% sits under both the Billo average (25.44%) and the Biddyco median (20.9%), so most viewers never reach what you're selling.
Swap in an opener that raises a question
Replace a product shot or logo open with a frame that doesn't explain itself: a raw photo, a problem in progress, or a line the viewer has to finish ("Do not buy this on Amazon"). Say the line out loud and burn it in as a caption so it lands with the sound off.
Try a native-looking first frame
If a polished open keeps losing, test one that reads as a post, not an ad: a comment-reply sticker, a text-message or Notes screenshot, a creator talking to a phone camera. Change the look of the first second before you rebuild the video.
Make 3–5 openers per video, not one
Cut every new video with several first-3-second variants on the same body. Launch them together and keep the one with the highest hook rate on the same placement; that's the only clean comparison.
Over 30% with a weak hold rate: fix the body
The hook is doing its job. Tighten the next 10 seconds instead: get to the product and the reason to buy faster.
Real ad · Carpe · 100 days in market
“Do NOT buy this on Amazon” — a hook built for the first 3 seconds
Carpe's 19-second video opens with the warning spoken over a close-up of the product and burned in as a caption, then pins the price bar for the full runtime. It ran 100 days (April 1 – July 10) and reached 576,305 UK accounts. Its hook rate isn't public, but the opener is the pattern this page grades: a line the viewer has to stop for, readable with the sound off.
Hook rate benchmarks by industry (2026).
The average Meta hook rate across industries is 25.44%, and the spread by category runs from about 21% to 29%. The largest industry-level dataset we could verify comes from Billo, a UGC-creator platform: 88,329 sales-objective Meta video ads with more than 1,000 impressions each, run January–June 2026 on $122 million of spend, with hook rate defined as 3-second video plays ÷ impressions. Its cross-industry average was 25.44%, up from 24.42% across 80,069 ads in H2 2025. Toys & Games (28.79%), Business & Industrial (28.74%) and Health & Beauty (28.11%) led; Animals & Pet Supplies (21.32%), Electronics (22.21%) and Baby & Toddler (22.56%) trailed. Read these as averages, not targets. Billo publishes means, not medians, and gives no placement or cold-versus-retargeting split. A category can also move almost four points in a half-year: Services went from 19.95% to 23.86%. Benchmark against your category, then against your own previous hooks.
| Industry | H1 2026 avg | H2 2025 avg |
|---|---|---|
| Toys & Games | 28.79% | 28.23% |
| Business & Industrial | 28.74% | 25.69% |
| Health & Beauty | 28.11% | 28.34% |
| Sporting Goods | 26.89% | 22.99% |
| Food & Beverages | 26.14% | 25.47% |
| Home & Garden | 25.06% | 24.47% |
| Services | 23.86% | 19.95% |
| Arts & Entertainment | 23.02% | 25.07% |
| Apparel & Accessories | 22.94% | 23.44% |
| Software | 22.90% | 22.78% |
| Baby & Toddler | 22.56% | 24.52% |
| Electronics | 22.21% | 25.00% |
| Animals & Pet Supplies | 21.32% | 19.80% |
| Other (small industries pooled) | 23.59% | 26.10% |
| All industries | 25.44% | 24.42% |
Source: Billo, "Video Ad Hook Rate Benchmarks H1 2026" (published August 6, 2026; 88,329 ads, Jan 1–Jun 30, 2026) and "Video ad hook rate benchmarks 2025" (published May 27, 2026; 80,069+ ads, Jul 1–Dec 31, 2025). Both cover Meta video ads with a sales objective and 1,000+ impressions. Billo notes that smaller categories (Baby & Toddler, Software, Services) swing more with individual advertisers. One smaller, independent dataset points the same way at the top end. Across 403 ads from 26 accounts of one direct-response info-product business (Q1 2024), Funnel Insiders put 38% as "a strong number to aim for." It also found no correlation between thumb-stop ratio and CPA or ROAS, a reminder that hook rate is a diagnostic, not the goal. Hook rate itself isn't in Meta's list of video ad metrics; it's built as a custom metric from Meta's 3-second video plays, which counts plays of at least three seconds and excludes replays.
- Billo — Hook Rate Benchmarks H1 2026
- Billo — Hook rate benchmarks 2025 (H2)
- Biddyco — Hook rate and hold rate benchmarks for Meta ads (2026; 479 ads, 17 US accounts)
- Funnel Insiders — thumb-stop ratio analysis (2024)
- Meta Business Help — video ad metrics (no hook rate in the list)
- Meta Business Help — 3-second video plays
- Meta Business Help — custom metrics
Hook rate by format: video, static and carousel.
Video
This is the only format hook rate exists for. Meta counts 3-second plays per impression and excludes replays. One quirk: a video shorter than 3 seconds counts as a play at about 97% of its length, so very short clips can read high.
Static images
No hook rate. A static ad has impressions but no video plays, so the formula has nothing to count and the column stays empty. Judge statics on CTR, cost per result and how long they keep spending instead.
Carousel
Meta sums the video metrics across every video card in a carousel, so its hook rate isn't comparable with a single-video ad. Read the cards one by one.
That gap matters more than it sounds. On our skincare ads page, 8 of the 10 longest-running ads are static images, though statics are 27 of the 60 ads (Meta Ad Library, as of November 1, 2026). Hook rate can only rank the videos, so it can't pick the winner on its own.
See the longest-running skincare ads →Hook rate by placement: Reels vs feed vs Stories.
There's no reliable published answer to whether Reels hooks less than Feed. Guides we checked in 2026 disagree on the direction: some say Reels run 5–10 points below Feed, others that Reels beat Feed by 5–8 points. None of them cites a dataset, and the large industry benchmark above has no placement split. So read it off your own account. In Ads Manager, keep your hook-rate custom metric in the columns, open Breakdown, choose placement, and compare the same ad across Facebook Feed, Instagram Feed, Reels and Stories. Your ad's overall hook rate is an impression-weighted blend of those placements. When Advantage+ placements shifts spend from one surface to another, the blended number moves even though the hook hasn't changed. Compare hooks within one placement, not across a mix that keeps changing.
Hook rate on TikTok vs Facebook: 2-second vs 3-second windows.
A TikTok hook rate isn't directly comparable to a Meta hook rate, because the two use different counting windows. On Meta, hook rate is conventionally built from 3-second video plays ÷ impressions. On TikTok, buyers usually build it from a 2-second view count ÷ impressions. A shorter window always counts more viewers, so the same creative reads higher on TikTok before the hook itself makes any difference. We haven't found a published, sample-disclosed dataset that measures both platforms on the same window, so we don't quote a TikTok-versus-Meta gap. To compare your own accounts, build a second Meta custom metric from 2-second continuous video plays ÷ impressions (Meta lists that metric among its video ad metrics, checked November 1, 2026) and set it beside your TikTok 2-second rate. Or skip the cross-platform comparison and judge each platform only against its own history. The window matters more than the platform name when you're reading someone else's number.
Thumb-stop ratio vs hook rate
"Thumb-stop ratio" is the same idea under an older name, and it's used for both windows: some guides define it as 3-second plays ÷ impressions, others as 2-second views ÷ impressions. Before comparing a quoted thumb-stop number with your hook rate, check which window it uses.
Hook rate vs hold rate vs thumb-stop rate.
Hook rate measures the first 3 seconds — did the hook stop the scroll. Hold rate measures whether they keep watching after that — did the body hold attention. A high hook rate with a low hold rate is a great hook on a video that doesn't deliver; a low hook rate caps everything downstream, which is why the hook is the first thing to fix.
Two hold-rate formulas circulate, so check which one a number you're quoted uses (the worked example above shows both). The funnel version many Meta buyers prefer (ThruPlays ÷ 3-second plays) isolates the body of the video from the hook, so a weak hook doesn't drag hold rate down with it. Either works; pick one and stay consistent.
Hook rate
3-second video plays ÷ impressions
Did the opener stop the scroll?
Thumb-stop rate
3-second plays ÷ impressions, or 2-second views ÷ impressions
The same question under an older name; check which window a quoted number uses.
Hold rate
ThruPlays ÷ 3-second plays, or ThruPlays ÷ impressions
Did the body keep the people the hook stopped?
What a high hook rate looks like.
The fastest way to internalise the metric is to look at hooks built to maximise it. Two from our teardown library, with the mechanism spelled out:
Why is my hook rate low? Five causes, in order.
A low hook rate is usually one of five things, and the cheap ones to rule out come first. Before you rewrite anything, check that "low" is measured against the right bar: your category's average in the table above, and your own best hook on the same account. Not a number quoted from another platform or another window.
The counting window
Confirm your custom metric divides 3-second video plays by impressions. A 3-second Meta number will always look low next to a 2-second TikTok or "thumb-stop" benchmark.
The placement mix
Break the ad down by placement. If one surface is dragging the blend, the hook may be fine and the delivery mix is what changed.
Audience temperature
Compare prospecting and retargeting ad sets separately. People who already know you stop more readily, so a cold-traffic hook rate reads lower for the same creative.
The first frame and first line
Watch the first second with the sound off. If nothing in the frame raises a question or the frame reads as an ad, that's the fix. Test 3–5 new openers on the same body, and change only the first 3 seconds.
The format
Polished brand spots tend to lose the first second to native-looking openers — UGC, screenshots, text-message or app-UI disguises. Try one native-format opener before you rebuild the whole video.
One caution before you chase the number: a low hook rate isn't always the problem. In one published dataset of 403 ads, thumb-stop ratio showed no correlation with CPA or ROAS. If the ad is converting at your target cost, the hook is doing enough.
How to improve your hook rate.
Hook rate is a hook problem, so fix the first 3 seconds: test 3–5 hook variants on the same video, change only the opener, and keep the winner. Use the library for the patterns and the generator to spin up variants fast. If new hooks keep losing, the angle underneath may be the problem, not the opener; and run the variants inside a proper creative test so the comparison is clean.
Hook rate questions.
Hook rate is the percentage of impressions that watch the first 3 seconds of your video ad — 3-second video plays ÷ impressions × 100. It measures how scroll-stopping your hook is.
A good hook rate for a Meta video ad is 25–35%. Billo's H1 2026 dataset of 88,329 sales-objective Meta video ads averaged 25.44% (cold and warm traffic mixed); 30% and above is strong, over 40% is elite, and under 20% means the opener needs replacing. Meta publishes no target, so after the band, compare your own hooks against each other on the same placement and audience.
About 21%. Biddyco's September 2026 analysis of 479 Meta video ads across 17 US ad accounts ($10.1M of spend, June–August 2026, each ad with at least $1,000 spent) found a median hook rate of 20.9%, with the middle half of ads between 14.2% and 30.4% and the top 10% at 44.8% or higher; consumer brands ran a 22.0% median and apps and services 20.0%. Its median hold rate (ThruPlays ÷ impressions) was 4.6%. Medians sit below Billo's 25.44% average because a few very high hook rates pull averages up.
Hook rate = 3-second video plays ÷ impressions × 100. Example: 10,000 3-second plays on 40,000 impressions is a 25% hook rate. Some teams use 2-second (thumb-stop) views instead of 3-second; stay consistent.
Hook rate measures the first 3 seconds (did the hook stop the scroll); hold rate measures whether viewers keep watching after that (did the body hold them). You want both high.
Change the first 3 seconds. Test 3–5 hook variants on the same video, swap only the opener, and keep the one with the highest hook rate. Retire hooks as they fatigue.
25–35%, with 30%+ strong and under 20% weak. Meta doesn't report hook rate or set a target for it; you build it as a custom metric from 3-second video plays ÷ impressions. The largest public dataset we could verify, Billo's H1 2026 benchmark of 88,329 sales-objective Meta video ads, averaged 25.44%, with no split by placement or by cold versus retargeting traffic. Placements can read several points apart for the same ad, so compare against your own hooks on the same placement.
In Billo's H1 2026 dataset of 88,329 sales-objective Meta video ads, the cross-industry average hook rate was 25.44%. Toys & Games (28.79%), Business & Industrial (28.74%) and Health & Beauty (28.11%) were highest; Animals & Pet Supplies (21.32%) was lowest. These are means, with no placement split, so compare against your own hooks too.
On Meta's 3-second basis, published guides put a good hook rate at 25–35% (Exposure) or 30–39% (Hawky), and call 40% and above elite. The typical ad sits lower: Billo's 88,329-ad H1 2026 average is 25.44% and Biddyco's 479-ad median is 20.9%. None of these is a Meta number, so start from your own account's median and try to beat it.
Hook rate is 3-second video plays ÷ impressions: did the opener stop the scroll. Thumb-stop rate is the same idea under an older name, sometimes measured on 2-second views instead of 3. Hold rate is ThruPlays ÷ 3-second plays (or ÷ impressions): did the body keep people. With 40,000 impressions, 10,000 3-second plays and 1,500 ThruPlays, hook rate is 25%, funnel hold rate 15% and per-impression hold rate 3.75%.
No. Hook rate is built from 3-second video plays, and a static image has no video plays, so there's nothing to count. Judge statics on CTR, cost per result and how long they keep running. It matters: on our skincare ads page, 8 of the 10 longest-running ads are static images (Meta Ad Library, as of November 1, 2026).
No — Ads Manager doesn't ship a hook-rate column. Build it once as a custom metric: 3-second video plays ÷ impressions, formatted as a percentage, then add it to your columns.
Found a competitor's video whose hook keeps stopping the scroll? Keep its hook, pacing and structure and put your product in every frame.
Put your product in a winning video ad →Low hook rate? The hooks winning in your niche already exist.
Paste your product link. ViralMojo finds the ads winning in your niche, works out why their first 3 seconds stop the scroll, and turns them into your own — static ads, AI video, a winning ad with your product in it, or a real creator who shoots it for you.
See the hooks winning in your nichePlans from $99/mo, flat — see ViralMojo pricing.