Creative Metrics That Matter: Hook Rate, Hold Rate and More

Hook rate is the first number most creative strategists check on a video ad, because it answers a simple question: did the opening stop the scroll? It is not the only creative metric that matters, though. Hold rate, average watch time, ThruPlays, click-through rate, CPM, cost per result and frequency each tell you something different, and they only make sense read together. This guide defines each one plainly, gives the formula, and shows how to calculate it with the video metrics Meta, TikTok and Snapchat actually report.

If you are new to testing video, our guide to AI UGC video ads uses the same definitions of hook rate and hold rate, so the two articles can be read side by side.

Why creative metrics matter

Creative metrics matter because the ad itself is now the main lever most advertisers still control by hand. Audience and bidding are increasingly left to each platform's delivery system, so the difference between a profitable week and a wasted one often comes down to which video ran. A campaign-level ROAS figure hides that. One strong creative can carry five weak ones, and the account average will look fine until the strong one tires.

Creative metrics split the journey into stages: attention (hook rate), interest (hold rate and watch time), action (CTR) and outcome (cost per result). When an ad underperforms, the stage where the numbers drop tells you what to change. A low hook rate points at the first line and first frame. A good hook with a weak hold rate points at the body. Strong attention with a poor click rate points at the offer or the call to action.

Hook rate and thumbstop ratio

Hook rate measures whether the opening of a video ad stops people scrolling. It is the share of impressions that turned into a short view.

Formula (Meta): hook rate = 3-second video plays ÷ impressions

Many teams call the same ratio the thumbstop ratio. Others use "thumbstop" for the raw number of short views and "hook rate" for the percentage, so always check which one a report means. Meta's developer documentation describes the 3-second event as a video watched for an aggregate of at least 3 seconds (Meta Marketing API, video ads guide).

What hook rate tells you: whether the first visual, the first line of copy or voice-over and the on-screen text earn a second look. What it does not tell you: whether the ad sells. A shocking opening can lift hook rate and attract the wrong people, so never judge a creative on hook rate alone.

Hold rate and average watch time

Hold rate measures how many people who got past the hook kept watching. It isolates the body of the ad from the opening.

Formula (Meta): hold rate = ThruPlays ÷ 3-second video plays, or 15-second video plays ÷ 3-second video plays

A ThruPlay on Meta counts a play to completion or to at least 15 seconds, whichever comes first, and the Marketing API documents a separate 15-second event for video watched for an aggregate of at least 15 seconds (Meta Marketing API, video ads guide). Because ThruPlay counts a short video when it finishes, a 9-second ad and a 30-second ad are measured on different terms. If you test ads of very different lengths, use a quartile metric such as plays at 50% or 75% instead, so each ad is judged on its own length.

Average watch time is the mean time viewers spent on the video. It is useful as a tiebreaker between two ads with similar hold rates, and for spotting where a long ad loses people. Look at it next to the video length, because 6 seconds of watch time is excellent on an 8-second ad and weak on a 40-second one.

Click and cost metrics

Click and cost metrics show whether attention turns into action at a price you can afford. Four of them matter most for creative work.

• CTR (link) vs CTR (all). Meta reports both. Link click-through rate counts clicks that go to your destination. CTR (all) also counts clicks on the profile, the "see more" text, reactions and other parts of the ad. For creative decisions, use link CTR: link clicks ÷ impressions. CTR (all) can rise simply because people expanded the caption.
• CPM. Cost per 1,000 impressions: spend ÷ impressions × 1,000. CPM is mostly an auction and audience signal, but a creative that the platform rates as relevant can win cheaper impressions, and a rising CPM on one ad is a common sign of fatigue.
• Cost per result. Spend ÷ results, where the result is the event you optimise for: a purchase, a lead, an install. This is the number that pays the bills, and the one every other metric should explain.
• Frequency. Impressions ÷ reach, the average number of times one person saw the ad. Frequency on its own is not bad. Frequency rising while CTR falls is.

Creative fatigue signals

Creative fatigue is the slow decline of an ad as the same audience sees it too often. It rarely shows up as one sudden drop; it shows up as several metrics drifting at once, compared with the ad's own early days rather than with other ads.

Watch for this pattern:

1. Frequency climbs week on week.
2. Hook rate and link CTR slide below the ad's first-week level.
3. CPM and cost per click rise for the same audience.
4. Cost per result follows a few days later.

By the time step four arrives, budget has already been wasted. Our article on what creative fatigue costs you puts numbers on that waste, and rotating creative before fatigue kills the account covers how to keep fresh variants ready so a tired ad can be swapped out early.

Calculating them on each platform

Each platform reports its own set of video metrics, so the same idea needs a slightly different formula on Meta, TikTok and Snapchat. The table below maps each creative metric to the closest native field. Labels in Ads Manager can differ slightly from API field names, so match by definition, not by name.

TikTok. TikTok's help article on video play metrics lists 17 metrics as of October 2026. A 2-second video view counts a play of at least 2 seconds within one impression session, excluding replays. A 6-second video view counts a play of at least 6 seconds, a full play of a shorter video, or at least one engagement in the first 6 seconds (TikTok Ads Manager help, video play metrics). That engagement rule means a 6-second view is not purely a watch-time measure. TikTok also reports "Average play time per video view" and views at 25%, 50%, 75% and 100%.

Snapchat. Snapchat's Marketing API defines a video view as an impression with at least 2 seconds of consecutive watch time or a swipe up on the Top Snap, and offers a separate time-based version that excludes swipe ups (Snap Marketing API, measurement). Use the time-based version for hook rate, otherwise a swipe up inflates the ratio. Quartiles (25%, 50%, 75%), view completion and average screen time cover the rest.

Meta. Use 3-second video plays for the hook, ThruPlays or 15-second plays for the hold, and plays at 25%, 50%, 75% and 100% to see where viewers drop. If you run creative across several brands or clients, our guide to tracking client creative performance across accounts shows how to keep these formulas consistent between accounts.

Why platforms don't compare directly

Hook rate and hold rate are not directly comparable across platforms because each one counts a view differently, serves ads in a different feed and builds an impression in its own way. Treat a Meta hook rate and a TikTok hook rate as two different metrics that happen to share a name.

The main differences:

• Thresholds. Meta's usual hook uses 3 seconds; TikTok and Snapchat start at 2 seconds. A 2-second threshold will almost always produce a higher ratio than a 3-second one on the same video.
• What counts as a view. Snapchat's default video view includes swipe ups, and TikTok's 6-second view includes early engagement. Meta's 3-second play is a pure watch-time event.
• Impression rules. TikTok counts a new impression session when someone swipes away and back, which affects both the numerator and the denominator.
• Feed context. Full-screen vertical feeds with sound on behave differently from mixed feeds where many people scroll with sound off.

The practical rule: compare creatives against each other within one platform, and compare platforms on cost per result and on the business outcome. If you want one view of all of them, keep a written formula per platform and apply it the same way every week.

Using the metrics in a creative testing workflow

A creative testing workflow works best when each round answers one question and each question has one metric. Testing a new hook, body and offer in the same variant tells you that something worked, not what.

A simple sequence:

1. Hooks first. Keep the body and call to action fixed and change only the first seconds. Judge on hook rate, with link CTR as a check that the hook attracts the right people.
2. Bodies next. Pair the two best hooks with different bodies. Judge on hold rate and average watch time.
3. Offer and call to action. Once attention is solid, test the closing. Judge on link CTR and cost per result.
4. Scale and monitor. Move budget to the winner and watch frequency, CTR and CPM for fatigue.

Give each test a fixed budget and end date before it starts, and only call a winner once each variant has enough impressions for the difference to be more than noise. Our creative testing framework for Meta ads covers budgets, test structure and how long to run a round. For the production side of video, our pillar on creating Facebook video ads with AI walks through the full workflow.

Benchmarks and your own baseline

Public benchmarks for hook rate and hold rate exist, but they are a weak guide. They mix industries, placements, video lengths and audience temperatures, and many do not say which threshold or formula they used. A figure that is healthy for a cold-audience TikTok ad in fashion tells you little about a retargeting ad on Meta for software.

A better approach is to set your own baseline:

• Pick a formula per platform and write it down, using the table above.
• Take your last 20 to 30 creatives on each platform with meaningful spend and calculate each metric.
• Use the median as your baseline, not the average, so one viral outlier does not skew it.
• Split by format and funnel stage where you have the volume: short and long video, cold and warm audiences.
• Refresh the baseline every month or quarter, because seasons, offers and auctions move.

New variants are then judged against your median: a hook rate clearly above it earns more budget; one clearly below it is a candidate to cut. The question shifts from "is this good?" to "is this better than what we usually make?", which is the one that matters.

How Wevion fits

Wevion is an ad operations platform that connects Meta, Google, TikTok, Taboola, Snapchat and Outbrain through their official APIs, syncs them into one view about every 15 minutes, and adds a bulk launcher, rules you approve, profit tracking (in beta) and Wavo, the AI copilot, on top of that shared data.

For creative work, Creative Analytics shows spend, CTR, CPA, ROAS and purchases for every ad, with images and videos filtered apart. Each ad also gets a 0 to 100 fatigue score that weighs CTR decay, frequency, CPM and CPC against the ad's own first 7 days, which is the fatigue pattern described above turned into one number. You can pause a tired Meta ad in one click, or set an ad-level rule that pauses anything past your CPA or ROAS line. Hook rate and hold rate still come from each platform's video fields, using the formulas in this guide.

Wavo, the AI copilot, reads performance across your connected accounts and proposes changes, such as pausing a fatigued creative or moving budget to a winner, and nothing changes until you approve it. You can try it with a 14-day free trial on the Pro plan, no card needed.

Key takeaways

• Hook rate (thumbstop ratio) is 3-second video plays ÷ impressions on Meta; it measures the opening, not the sale.
• Hold rate is ThruPlays or 15-second plays ÷ 3-second plays on Meta, or 6-second ÷ 2-second views on TikTok.
• Use link CTR, not CTR (all), and read CPM, cost per result and frequency alongside the video metrics.
• Fatigue shows as frequency rising while hook rate and CTR fall against the ad's own first week.
• Thresholds differ by platform, so compare creatives within a platform and platforms on cost per result.
• Skip borrowed benchmarks: set a median baseline from your own recent creatives and refresh it regularly.