A good hook rate benchmark for Meta ads in 2026 is 25% or higher on cold traffic. 30%+ is good, 35% to 45% is the elite band scaled accounts live in, and anything under 20% needs work (under 15% is a kill). On TikTok, where the hook window is 2 seconds instead of 3, aim for 30% to 40%. Those are the headline numbers. The tables below break them out by audience, placement and vertical, and the second half of this page is a decision tree for what a bad number actually means.
One paragraph of housekeeping first. Hook rate is 3-second video plays divided by impressions. Meta counts a 3-second play when the video runs for 3 seconds, or 97% of its length if it is shorter, and replays are excluded, per Meta's own metric definition. Thumbstop rate is the same metric wearing a different name. If any of the vocabulary here is new, our creative metrics glossary pins down the whole family with formulas. This page does the other job: what counts as good, and what to do when your number is not.
What is a good hook rate for Meta ads in 2026?
The bands below are for a single creative on cold prospecting traffic, read after at least 2,000 impressions. Every row traces to a published source.
| Band | Hook rate (cold) | What it means | Source |
|---|---|---|---|
| Kill zone | Below 15% | Dead on arrival. Pull it and reshoot the open. | Ryze, 500M+ impression panel |
| Fix zone | 15% to 20% | The open cannot stop a scroll. Recut the first 3 seconds before spending more. | Spark interpolation, between Ryze's sub-15% kill line and Exposure's sub-20% "weak" cut |
| Workable | 20% to 25% | Par for most accounts. Acceptable while CPA holds. | Vaizle (floor for most accounts) |
| Good | 25% to 35% | The healthy cold-audience range in 2026. | Ryze |
| Elite | 35% to 45% | Retargeting-level attention from strangers. Scale candidate, pending hold rate. | Ryze's own retargeting band, used here as the cold-traffic stretch goal; Exposure's elite threshold starts at 40%+ (30%+ counts as strong) |
These bands broadly match the ones in our Meta ads benchmarks by business type: elite 35% to 45%, with anything below 20% needing work on established accounts.
The bands also now have a measured average to sit against. In August 2026 Billo published a benchmark panel of 88,329 sales-objective Meta video ads that ran between January and June 2026, covering $122 million in spend across 14 industries. The cross-industry average hook rate came out at 25.44%, up from 24.42% in the second half of 2025. So the average ad now lands at the bottom of the good band, which is exactly why 25% reads as a floor rather than a target.
Where to read it in Ads Manager
Meta has no stock column labelled "hook rate", though its Help Centre documents the underlying metric as "3-second video plays rate per impressions." Build it once as a custom metric: in Ads Manager, open Columns, choose Customise columns, then Create custom metric, and define it as 3-second video plays divided by impressions, formatted as a percentage. The denominator is impressions, never reach and never plays; swapping it quietly inflates the number and breaks every comparison against the tables on this page. In TikTok Ads Manager the equivalent is 2-second video views divided by impressions. Save both at account level so every report your team reads carries the same definition.
Context changes what you should expect from the number, so here is the same metric cut by audience and placement.
| Context | Expected hook rate | Source |
|---|---|---|
| Cold prospecting, Feed | 25% to 35% | Ryze |
| Retargeting | 35% to 45% | Ryze |
| Reels, cold DTC | 24% to 36%, median around 30%; the faster swipe means a Reels number must clear a higher bar to match Feed quality | AdSights |
| Stories, DTC | 22% to 32%, median around 27% | AdSights |
| Cold prospecting, observed range | 18% to 28% is where most accounts land, median around 22% | AdSights |
That last row matters more than it looks. If your cold prospecting number sits at 22%, you are sitting on AdSights' observed median, not failing; the five bands in the first table describe where you should aim, and Billo's measured 25.44% average confirms most ads land near the bottom of the good band. Note the direction of the Reels row too: a full-screen placement posting a bigger raw number than Feed is not automatically performing better, because the faster swipe inflates the numerator. Mixing placements and audiences into one blended read is the most common way operators scare themselves.
What is a good hook rate on TikTok?
TikTok counts the hook at 2 seconds, not 3. A shorter bar is easier to clear, so TikTok numbers run higher than Meta's for the same creative. The gap is mechanical rather than a sign the ad performs better there, which is why you benchmark within one platform and never across. There is no published TikTok-specific hook rate panel at Meta's scale; Billo's growth metrics guide sets a working target of 30% to 40%, with hold rate above 25%, and the table below reuses MHI Growth Engine's Meta DTC thumb-stop bands as a rough proxy in the same range.
| Band | Hook rate band (MHI's Meta DTC data, used as a TikTok proxy) | Source |
|---|---|---|
| Weak | Below 20% | MHI Growth Engine |
| Average | 20% to 30% | MHI Growth Engine |
| Good | 30% to 40% | MHI Growth Engine; Billo (working target) |
| Excellent | 40%+ | MHI Growth Engine |
Do hook rate benchmarks change by vertical?
Yes, and the data here got materially better in August 2026, when Billo's H1 2026 panel published measured averages for 14 industries. The measured numbers run lower than the round figures most guides quote, so treat any single-vertical claim without a panel behind it with some suspicion. Here is what is actually sourceable now.
| Vertical | What the data shows | Source |
|---|---|---|
| Health and beauty | 28.11% measured average in H1 2026; benchmark against 28%, not the 25.44% cross-industry figure. AdSights' observed range for DTC beauty is 25% to 35%. A swatch or routine is visually arresting at scroll speed, which is why the category sits above the line | Billo H1 2026 panel; AdSights |
| Toys and games | 28.79% measured average, the highest of the 14 industries in the panel; the satisfying demo and the unexpected reveal do the work | Billo H1 2026 panel |
| General DTC ecom | 25% to 35% on cold traffic | Ryze |
| B2B and high-consideration | Exposure only says a considered, high-AOV purchase "pulls different attention" than an impulse buy, without a discount figure; Billo measured Business and Industrial at 28.74% for the half, above the cross-industry average, so do not assume a discount before you have your own read | Exposure; Billo H1 2026 panel |
| Pet supplies | 21.32% measured average, below the cross-industry line in all six months of H1 2026 | Billo H1 2026 panel |
Beyond those rows, your own account average at a similar video length is the vertical benchmark. A 22% hook rate that replaces an 18% control is a win regardless of what the beauty brands are posting.
The bands tell you where you stand. The hook and hold pair tells you what to fix.
How do you diagnose a bad hook rate?
A hook rate on its own is a symptom, not a diagnosis. To know what to fix you need it next to hold rate (ThruPlays divided by 3-second plays; healthy is 20% to 25% on cold traffic per Ryze, and below 18% is a problem). Sources disagree on hold bands more than any other metric because half the industry uses a different formula; the glossary covers that mess so this page does not have to.
Rule zero before any branch: do not read hook rate under 2,000 impressions per creative, and expect it to settle between 5,000 and 10,000. The budget maths for getting there without torching your testing pot is in how much to spend per creative. From 2,000 impressions, work down this tree.
Low hook rate, healthy hold rate
The concept works. The open fails. People who make it past 3 seconds stay, which means the script, product and offer are doing their jobs and the first frame is not. This is the cheapest problem on the page: recut the first 3 seconds from footage you already have, engineer a pattern interrupt into the first frame, rewrite the opening line or drop a text overlay on it. You do not need a new shoot. Our library of UGC hooks that convert and the 7 hook angles framework exist for exactly this branch.
Healthy hook rate, low hold rate
The opposite problem, and the sneakier one. The open is bait the body cannot pay off. A shouted first line and a jump cut will buy you 3 seconds; they will not buy you 15. Keep the hook, rebuild seconds 4 to 15, and check that the promise in the first line is answered by second six. Chasing an even higher hook rate here wastes money on the half of the ad that already works.
Both low
Both numbers failing means the concept itself missed, and no amount of recutting rescues a missed concept. Do not let it keep spending while you deliberate. The spend caps in the 1.5x CPA kill rule put a hard ceiling on what a loser is allowed to burn. Brief a new angle rather than a new edit of the same one.
Both healthy, CTR flat
The ad is entertaining people for free. It earns attention and holds it, then asks for nothing. Look at the CTA, the offer and the end card before touching the hook, because the attention half of the funnel is already solved.
Hook rate was fine, now sliding
If the same creative posted 30% three weeks ago and posts 22% today with frequency creeping past 2.5, the hook did not get worse. The audience has seen it. That is fatigue, and the fix is cadence, not surgery on this one ad: the weekly creative quota by spend covers how many fresh tests keep an account ahead of it.
What we see inside client accounts
Spark is a done-for-you creative studio, so we spend most of our week inside other people's Ads Manager reports. In our view, when hook rates are stuck under 20%, the fix worth trying first is almost never new footage: recutting the first 3 seconds of existing ads into two or three new opens usually works, because the concept tends to survive review while the open does not. The pairing worth watching for in the other direction: a hook rate clearing 25% next to a collapsing hold rate almost always traces back to a first second that got all the attention while the promise at second four was neglected. We do not have a published panel of our own to hand you hard numbers from, and we will not invent one; the sources above are the citable data. The view stands regardless: diagnose the pair, and the expensive fix is usually the wrong first move.
Key takeaway
Meta cold traffic in 2026: below 15% kill, 20% to 25% workable, 25% to 35% healthy, 35% to 45% elite. The measured cross-industry average in H1 2026 was 25.44%, per Billo's 88,329-ad panel. TikTok (2-second window): 30% to 40% good. Read nothing before 2,000 impressions, and never diagnose a hook rate without hold rate sitting next to it.
If the tree keeps landing on "the open is the problem" and there is no footage left to recut, that is the gap we fill. Spark runs research, scripting, creators and iteration as one weekly loop, so new opens ship on a schedule instead of when someone finds time. If you are weighing outside help more broadly, our ranking of the best UGC agencies for Meta ads compares the options, including us, by use case. If your hook rates have been camped in the fix zone for a quarter, book a call and we will look at the account together.
Frequently asked questions
What is a good hook rate for Facebook ads?
On cold traffic in 2026, 25% or higher is healthy, 30%+ is good, and 35% to 45% is elite. Below 20% the creative needs a new opening, and below 15% most operators kill it. Retargeting audiences should run around 10 points higher than cold, because the viewer already knows the brand.
Is hook rate the same as thumbstop rate?
Yes. Thumbstop rate (or thumbstop ratio) is the same calculation under a different name: 3-second video plays divided by impressions on Meta. TikTok counts the hook at 2 seconds instead of 3, so the same creative posts a higher number there. Benchmark within one platform, never across platforms.
How many impressions do you need before hook rate is reliable?
As a rule of thumb, treat 2,000 impressions per creative as the minimum before reading hook rate, and expect the number to settle properly between 5,000 and 10,000 impressions. Reads taken earlier swing too much to act on, so a kill decision at 500 impressions is a coin flip dressed up as analysis.
What is a good hold rate?
Using Meta's ThruPlay definition (ThruPlays divided by 3-second plays), 20% to 25% is healthy on cold traffic and below 18% signals a body problem. Some guides, e.g. Vaizle, aim for 40% to 50% hold rate; benchmarks vary widely, so always write the formula next to the number in your reports.
Can a hook rate be too high?
Not by itself, but a high hook rate with a collapsing hold rate means the opening writes a cheque the ad never cashes. Shock openings and bait cuts buy the first 3 seconds and lose the viewer at second four. Judge the hook and hold pair together, never the hook rate alone.