Hook rate is the share of impressions where someone watched at least the first 3 seconds of your video ad: 3-second video plays divided by impressions. Hold rate is the share of those 3-second viewers who stayed to a ThruPlay, which Meta counts at 15 seconds or completion: ThruPlays divided by 3-second plays. And thumbstop rate is hook rate under a different name. That clears up 80% of the confusion already. The rest of this page pins down the whole family, with formulas, 2026 benchmarks and what to do when a number comes back low.
The reason a glossary needs to exist at all: nobody agrees on vocabulary. Motion says thumbstop ratio. Ads Manager says 3-second video plays and hides the ratio entirely unless you build a custom metric. Half the articles defining hold rate use a different denominator from the other half. When a strategist, a media buyer and a founder read the same report and mean different things by the same word, the creative debrief turns into a translation exercise. So: one page, every definition, sources linked.
What is hook rate?
Hook rate is the percentage of people who watched the first 3 seconds of your video ad. The formula: 3-second video plays ÷ impressions × 100. Serve 80,000 impressions, get 20,800 3-second plays, and your hook rate is 26%.
It answers exactly one question: does the opening stop the scroll? Nothing else. A strong hook rate says nothing about whether the ad sells, which is why judging creative on hook rate alone rewards loud openings over good ones.
Benchmarks: Ryze's 2026 analysis of 500M+ Meta impressions puts the cold-audience range at 25% to 35%, retargeting at 35% to 45%, and calls anything under 15% dead on arrival. Vaizle's guide sets a workable floor at 20% to 25% for most accounts. If you sit below your account average, fix the first 3 seconds and only the first 3 seconds: the visual open, the first spoken line, the text overlay. We keep a list of openings that reliably clear the bar in our breakdown of UGC hooks that convert.
What is thumbstop rate, and is it different from hook rate?
It is not different. Thumbstop rate (also thumbstop ratio, or TSR) measures the same thing as hook rate: the percentage of served impressions where the viewer watched the opening seconds. Motion's metrics cheat sheet defines it as the share of people watching the first 2 to 3 seconds of your video, with the window varying by platform. Meta counts 3 seconds. TikTok counts 2.
That one-second difference matters more than it looks. A 30% thumbstop on TikTok and a 30% hook rate on Meta are not the same achievement, so cross-platform creative comparisons on this metric quietly mislead. Pick one name for your internal reporting, note the platform window next to it, and move on.
What is hold rate?
Hold rate is the percentage of 3-second viewers who kept watching to a ThruPlay: ThruPlays ÷ 3-second video plays × 100. Where hook rate grades the opening, hold rate grades the script. It tells you whether the video kept the promise the hook made.
Benchmarks disagree here more than anywhere else in this glossary, and it is worth knowing why rather than pretending one number is canon. Ryze's panel calls 20% to 25% healthy on cold traffic, 30% to 40% on retargeting, and below 18% a body problem. Vaizle recommends aiming for 40% to 50%. Both are defensible, because a ThruPlay means different things at different lengths: for a 15-second video it is completion, for a 60-second video it is a quarter. Longer videos bleed hold rate by definition. Benchmark against your own account average at a similar video length, and treat published ranges as calibration.
One more trap. Some teams calculate hold rate as views-to-25%-of-video ÷ 3-second plays instead of ThruPlays ÷ 3-second plays. Neither is wrong. But write the formula in the report header, because a 55% hold on the 25%-view definition and a 22% hold on the ThruPlay definition can describe the same video.
How to act on it: if hook rate is fine and hold rate is not, keep the hook and rebuild seconds 4 to 15.
What is a ThruPlay?
A ThruPlay is Meta's own unit, not an industry nickname. Per Meta's Business Help Center, it counts when a video of 15 seconds or shorter is played to completion, or when someone watches at least 15 seconds of a longer one. Meta treats 97% of a video's length as complete, because viewers drop off during end fades. ThruPlay doubles as a bidding option on video views campaigns, so you can optimise and pay per ThruPlay rather than per impression.
ThruPlay rate (ThruPlays ÷ impressions) collapses hook and hold into one number: 18% to 28% is Ryze's 2026 cold-audience range, 25% to 35% for retargeting, and anything above 30% reads as a scale signal. Useful as a summary, weak as a diagnostic. A bad ThruPlay rate cannot tell you whether the hook or the body failed, which is the question you actually need answered.
What is cost per ThruPlay?
Cost per ThruPlay is spend ÷ ThruPlays: what you pay for each person who genuinely watched. Ryze's 2026 range runs $0.01 to $0.10, with ecommerce typically at $0.02 to $0.06 and anything above $0.15 worth investigating. Treat it as a primary KPI only when you are actually buying ThruPlays. On conversion campaigns it is a diagnostic, not a target: optimising a purchase campaign around cheap views is how you end up with entertained audiences and empty carts.
What is first frame retention?
First frame retention is Motion's name for the percentage of people who let your video start playing at all in their feed. TikTok's equivalent is thumbnail retention. It sits upstream of hook rate: before anyone can watch 3 seconds, the first frame has to survive the flick of a thumb. Motion's guidance is blunt: below 90%, change the first frame. Test a text overlay that says what the ad is about, or swap a product-first open for a face-first one. This is the cheapest fix in the entire stack, because it is one frame.
What is thumbstop CTR?
Thumbstop CTR is the percentage of people who watched the first 3 seconds of your video and then clicked through to your site: link clicks ÷ 3-second video plays. Motion pairs it with thumbstop ratio to separate attention from action, because an ad can be brilliant at stopping thumbs and useless at moving them. A wide gap between thumbstop CTR and plain link CTR means people are clicking without really watching, which shows up a lot in retargeting where the audience already knows the product. There is no credible public benchmark for this one, so compare against your account average and resist inventing a target.
Where does CTR fit?
CTR (link clicks ÷ impressions) is not an attention metric, and that is exactly why it belongs here: it is the sanity check on all of them. Every attention number on this page can glow green while CTR sits flat, and when that happens you have made content, not an ad. For DTC feed placements, around 1% is workable and 2%+ is strong. Our Meta ads benchmarks by business type break CTR out by vertical if you want a target closer to your category.
The benchmark table
Cold-audience Meta placements unless noted. Ranges from Ryze's 500M+ impression panel, Vaizle and Motion, linked above.
| Metric | Formula | Decent | Strong |
|---|---|---|---|
| Hook rate | 3-second plays ÷ impressions | 20% to 25% | 30%+ (kill below 15%) |
| Thumbstop ratio | Same as hook rate (Meta 3s window, TikTok 2s) | 20% to 25% | 30%+ |
| Hold rate | ThruPlays ÷ 3-second plays | 20% to 25% cold | 30% to 40%+ (worry below 18%) |
| ThruPlay rate | ThruPlays ÷ impressions | 18% to 28% cold | 30%+ (scale signal) |
| First frame retention | Videos that start playing ÷ impressions | Above 90% | A floor, not a ladder: fix below 90% |
| Thumbstop CTR | Link clicks ÷ 3-second plays | Vs account average | Vs account average |
| Cost per ThruPlay | Spend ÷ ThruPlays | $0.02 to $0.10 | Under $0.02 (investigate above $0.15) |
| CTR | Link clicks ÷ impressions | 1% to 2% | 2%+ |
Key takeaway
Every metric in this glossary answers one of two questions. Did they stop: first frame retention, hook rate, thumbstop ratio. Did they stay: hold rate, ThruPlay rate. Published benchmarks calibrate your expectations. Your account average delivers the verdict.
How we read these numbers at Spark
When a new creative batch goes live, we read the metrics in a fixed order, and the order matters more than the numbers. Hook rate first, at roughly 2,000 impressions per creative; earlier than that the figure jumps around too much to trust. It grades the first 3 seconds and nothing else. Hold rate second, and this is the one we actually trust to predict a scaler. Hook rate is the easiest metric on this page to game. A jump cut and a shouted first line will buy you 3 seconds. They will not buy you 15. When a creative holds attention all the way through the pitch, something real is happening. CTR third, to confirm attention is turning into intent. CPA last, once spend has earned a verdict, which takes longer than most people want it to; we covered the exact budgets in how much to spend per creative.
Hook rate tells you the ad got seen. Hold rate tells you it deserved to be.
The combination is the diagnosis. High hook and high hold: scale candidate. High hook and weak hold: keep the hook, rebuild the body; the script gets rewritten, not the opening. Weak hook and strong hold: the opposite, reshoot the first 3 seconds and leave the script alone. Weak on both: kill it and move on, using the spend caps in our 1.5x CPA kill rule so a loser never burns more than it should. Billo's growth-metrics guide lands on the same pairing logic: read hook and hold together, or you will fix the wrong half of the ad.
That is the whole glossary. Bookmark it, build the two custom metrics from the FAQ below, and your next creative debrief gets shorter. There are more references like this in our resources library. And if you would rather have the reading done for you, every Spark batch ships with hook and hold analysis built into the iteration loop at flat monthly pricing. Book a call and we will walk your last batch through this exact sequence.
FAQ
Why doesn't Meta Ads Manager show hook rate or hold rate by default?
Because they are custom metrics, not standard columns. In Ads Manager go to Columns, then Customise columns, then Create custom metric. Build hook rate as 3-second video plays divided by impressions, and hold rate as ThruPlays divided by 3-second video plays. Both take about two minutes to set up and then appear as normal reporting columns.
Are Meta and TikTok attention metrics comparable?
Not directly. Meta counts a hook at 3 seconds while TikTok counts 2, and TikTok's version of hold rate is measured at 6 seconds or a completed play. The same creative will post different numbers on each platform, so benchmark within a platform and never across.
Which metric should decide when to kill an ad?
Hook rate is the earliest reliable kill signal: below roughly 15% on cold traffic after about 2,000 impressions, recovery is rare. Spending decisions should still run on CPA rules, because an ad with a mediocre hook rate can occasionally convert well enough to earn its place.
Do these benchmarks change by industry?
Yes, a lot. The ranges in this glossary are cross-vertical: a 22% thumbstop ratio in B2B SaaS can be genuinely strong while the same number in beauty or food is below par. Use published ranges to calibrate expectations and your own account average to make decisions.