Mobile game ad hooks: test the first 3 seconds
The opening seconds carry a large share of a video ad's effect and are, we think, the cheapest part to test. Hook types game ads use, where a hook turns misleading, and how to run and read a hook test.
The first seconds of a video ad carry a large share of its effect. Facebook's (now Meta's) 2016 analysis with Nielsen found up to 47% of a video campaign's value was delivered in the first 3 seconds. A 2021 TikTok meta-analysis put 50% of an ad's impact in the first 2. So the hook is the part of a game ad to test most. We think it's also the cheapest part to test: swap the opening, lock the body, and every result reads as a verdict on the hook alone.
Why the first seconds carry the ad#
Most of what a video ad does, it does early. In February 2016 Facebook reported, with Nielsen, that "up to 47% of the value in a video campaign was delivered in the first three seconds". Up to 74% came in the first ten. The same post found that 65% of people who watch the first 3 seconds go on to watch at least 10, and 45% continue to 30.
TikTok's numbers point the same way. A 2021 meta-analysis with MediaScience, published by TikTok in 2023, found that "50% of the impact from a TikTok ad is realized in the first 2 seconds". The first 6 seconds captured 90% of the cumulative impact on ad recall.
| Meta 2016: value in the first 3 s | 47% |
|---|---|
| Meta 2016: value in the first 10 s | 74% |
| TikTok 2021: impact in the first 2 s | 50% |
| TikTok 2021: recall impact in the first 6 s | 90% |
Both studies are old and measure different things, so read each pair on its own. The direction has held. Amplified Intelligence, an attention-measurement vendor, reported in 2025 that social formats get 0.4–1.6 seconds of active attention per view, most of it in the first 2.5 seconds. Back in 2016, Facebook IQ measured 1.7 seconds spent with a piece of mobile feed content, against 2.5 on desktop.
The networks write their guidance around the same window. TikTok's creative best practices say to "introduce your content proposition in the first 3 seconds" and "prioritize your hook in the first 6 seconds". Google's App campaign guidance asks you to "grab their interest within the first 2 to 3 seconds", and notes that successful app video ads have "2+ cut changes within the first 5 seconds".
For games, this covers most of what you ship. SocialPeta and Reforged Labs, two ad-intelligence vendors, counted video at 78.1% of game creatives on Android and 73.9% on iOS (via GameDev Reports, December 2025). The opening is the most-watched part of the most-used format.
What a hook has to do depends on the placement#
Where the ad runs changes what the opening has to do.
In a feed, the viewer can scroll away at any moment. The hook's job is to stop the thumb, often with the sound off. Facebook's 2016 post found "41% of videos were basically meaningless without sound", so a feed opening should read from the picture alone.
In an interstitial, the viewer can't leave straight away. AppLovin says interstitials run at least 5 seconds before the skip. Its creative guide says a strong hook is what keeps users "engaged past the 5-second unskippable window" (AppLovin). Here the opening doesn't need to stop anyone. It needs to stop them reaching for the skip button.
Rewarded video is a third case. AppLovin says it can run unskippable for up to 60 seconds, and the viewer chose to watch it for the reward. The opening isn't fighting a thumb, so we'd put less weight on the first 3 seconds there and more on whether the game is clear by the end card.
| Feed | Interstitial | Rewarded | |
|---|---|---|---|
| What the viewer can do | Scroll away at any moment | Skip after at least 5 seconds | Watch for a reward, up to 60 seconds unskippable |
| The hook's job | Stop the thumb, often with sound off | Hold them past the skip button | Make the game clear before the end card |
| What to read first | Hook rate | Viewers still there after the skip | Completion, then installs |
One concept can run in all three. Test its hooks per placement, though: a feed winner and an interstitial winner answer different questions.
Hook types game ads use#
No primary data says which hooks game ads use most, or which work best. What exists is a working vocabulary UA teams share, and a little directional data from TikTok. These are the openings you'll see across mobile game ads:
- Gameplay first. The first frame is the game, mid-action. Unity's video best practices say to "front load your video with the best content" and to prioritise authentic gameplay.
- Fail and rescue. A player makes the wrong move and the viewer wants to fix it. Royal Match's "Save the King" format, called out in the same SocialPeta report, is one example.
- Character reveal. The hero or the villain fills the frame first, and the game follows.
- Creator or player reaction. A person reacts before the viewer sees the game. In a 2025 study of 887 ads by System1, a research vendor, with TikTok, not games-specific, creator-led ads got 39% more attention (Chief Marketer).
- Progression, before and after. A bare base becomes a fortress, a level 1 hero becomes level 50. The opening promises the change.
- Satisfying loops. Sorting, merging, cutting, filling: motion that is pleasant to watch on its own.
The closest thing to a ranking is TikTok's hyper-casual game creative tips (June 2023). They advise a hook in the "first 3-6 seconds" and report directional click-through lifts from December 2022: highlighting failure +9.99%, sound effects +7.61% and "relaxing" gameplay +7.71%. TikTok doesn't state the baseline, and the data covers one genre on one network. Treat these as hypotheses for your own test, not a ranking.
Where a hook becomes a misleading ad#
A hook can win the click by promising a game you don't ship. The UK regulator has ruled on exactly that.
In September 2020 the Advertising Standards Authority upheld complaints against Playrix ads for Homescapes and Gardenscapes. The ads showed pin-pull puzzles under lines like "ONLY 5% CAN SOLVE THIS". Playrix said those puzzles appeared only "10 times in total" in the game (Game Developer). The ads carried a disclaimer, and it didn't save them.
The ASA's advice on mobile game ads, updated in May 2025, says they should be "representative of the gameplay". A disclaimer is "unlikely to prevent an ad from misleading". That's the UK position; check the rules in each market you run.
Players notice too. In Mistplay's 2025 loyalty survey of 4,500+ gamers, run with AppsFlyer, 36% cited "ad creative matching gameplay" as a factor in choosing a game (Mistplay, a vendor survey). We'd expect an opening that attracts people wanting a different game to show up later as weaker early retention. A hook test that reads retention is where you catch it. Our post on why AI game ads look generated covers the same problem from the art side: the ad has to look like the game.
Show the brand early, without a logo card#
Early branding used to cost views. A 2015 Think with Google analysis of YouTube ads, on data from 2012–14, found that showing the brand in the first 5 seconds raised recall and awareness, but made people "more likely to skip". Floating logos hurt.
Newer studies split the brand from the logo. VCCP Media, an agency, and the attention researcher Karen Nelson-Field reported in 2025 that 1.5 seconds of genuine attention is enough to encode memory. Well-branded ads were 2.5x more effective in low-attention settings (72 ads, compared with "bad twins"). The System1 study with TikTok found that logo overlays reduced the awareness lift.
The findings agree once you separate the brand from the logo. A logo card interrupts. The game's own character, board and art style are branding that is also the hook. So let the first frame carry the brand without a logo: the hero, the world, the colours players will meet in the store. Save the logo for the end card.
Where the hook text sits matters as much. Meta asks you to keep the bottom 35% of a Reels or Stories frame clear of text, so a headline placed there competes with the app's own captions and buttons. Our guide to mobile game ad sizes has the safe zones.
How to measure a hook#
Two ratios do most of the work. The common convention, as Motion sets it out:
- Hook rate = 3-second video plays ÷ impressions. Some teams call it thumb-stop ratio.
- Hold rate = 15-second plays ÷ 3-second plays. On Meta, ThruPlays are often used instead; they also count shorter videos watched to the end, so for a sub-15-second ad the hold rate becomes a completion rate.
Hook rate says whether the opening stopped people. Hold rate says whether the body kept the ones it stopped. Neither says whether they installed.
Benchmarks are thin. The one average we found with a method behind it is Billo's: 24.42% across 80,069 Meta video ads in the second half of 2025, with Toys & Games at 28.23%. Billo is a vendor, and those were sales-objective campaigns for e-commerce, not app installs. We found no method-disclosed hook-rate benchmark for game app-install campaigns. The "top 10%" and "elite" hook rates that circulate don't say where they come from. So don't set a target from someone else's average. Compare each new opening with your own control opening, on the same body, in the same test.
Then read past the hook. Check each opening's cost per install and installs per thousand impressions, then its D1 and D7 retention. If a new opening raises hook rate and lowers CPI while D7 falls, our read is that it found cheaper players who wanted a different game. Measure in your ad manager or attribution tool; Vidscape makes the creatives and doesn't read campaign results.
Read hook rate and hold rate together#
The two ratios are most useful side by side. Here is a simple 2×2 for deciding what to change next. It's our framework, not a benchmark.
- HOOK HIGH · HOLD LOWFix the bodyThe opening stops people and what follows loses them. Keep the hook and test the next 10 seconds.
- HOOK HIGH · HOLD HIGHScale itThe opening stops people and the body keeps them. Check CPI and early retention before you add spend.
- HOOK LOW · HOLD LOWRethink the conceptNeither half lands. A new opening rarely rescues it, so spend the next test on a new idea.
- HOOK LOW · HOLD HIGHFix the openingThe few who get past 3 seconds stay. Lock the body and test new openings against it.
High hook, low hold is the case that's easy to misread. A strong opening in front of a slow body looks like a failed concept, when the fix is the next 10 seconds. Low hook, high hold is what a hook test is for. The body works, and a better opening will bring more people to it, as long as the audience that stayed isn't a small, self-selected one.
How to run a hook test#
A hook test changes one thing, the opening. Everything after it stays the same.
- Lock the body. Pick a control ad, ideally a current winner. Freeze everything after the first 3 seconds: footage, voice-over, end card.
- Write 3–5 openings. Change the first shot, the first line on screen or the character, with one idea per opening. Make them different enough to be real alternatives.
- Run a controlled test. Meta groups ads that share key visual and thematic attributes, and shares learnings and delivery at the creative level (December 2025). Our read is that near-identical ads dropped into one ad set may not get separate delivery. So use an A/B test or the network's creative-testing tool, with the control in it.
- Read the hook, then the money. Hook rate and hold rate against the control first, then CPI, IPM and D1/D7.
- Keep the winner as the new control, and start the next round from it.
- Lock the bodya control ad
- Write 3–5 openingsone idea each
- Run a controlled testcontrol included
- Read hook and holdthen CPI and D1/D7
- Keep the winnerthe new control
The loop runs as fast as you can make a new opening. In our experience, built by hand, a new first 3 seconds usually means a new edit and a new export in every size. In Vidscape a motion-graphic ad is built from editable layers, so a different character or headline in the opening is a variation of the same ad. In a video ad cut from generated footage, the hook text and end card stay editable and a single scene can be re-run. The page for growth teams covers the testing side.
Frequently asked questions#
What is a good hook rate for a mobile game ad?
There's no method-disclosed benchmark for game app-install campaigns. The one average we found with a method, Billo's 24.42% across 80,069 Meta video ads (second half of 2025), comes from e-commerce sales campaigns. Judge a new opening against your own control opening, on the same body.
Is hook rate the same as thumb-stop ratio?
Usually, yes. Both names describe 3-second video plays divided by impressions. Check how your reporting tool defines it before you compare numbers across tools.
How long should a mobile game ad hook be?
The networks put it in the first 2–6 seconds. Google asks App campaign videos to grab interest within 2 to 3 seconds; TikTok asks for the proposition in 3 and the hook within 6. In interstitials, plan for at least 5 seconds before the skip button appears (AppLovin).
Does a disclaimer make an unrepresentative game ad acceptable?
Probably not in the UK. The ASA says game ads should be representative of the gameplay, and that a disclaimer is unlikely to stop an ad from misleading. Its 2020 ruling against Playrix's Homescapes and Gardenscapes ads upheld complaints despite one.
- Updated features for video ads — Meta for Business · Feb 2016
- Capturing attention in feed: the science behind effective video creative — Facebook IQ · Apr 2016
- Resonance: a key factor in ad effectiveness (TikTok × MediaScience meta-analysis, 2021) — TikTok for Business · Sep 2023
- Creative best practices — TikTok Business Help Center · Jun 2025
- TikTok hyper-casual game creative tips — TikTok Creative Center · Jun 2023
- App campaign asset optimisation guidance — Google Ads Help
- Creating YouTube ads that break through in a skippable world — Think with Google · Jun 2015
- Designing creatives — AppLovin · Feb 2026
- High-performing video creatives — AppLovin
- Video creative best practices — Unity
- SocialPeta and Reforged Labs: mobile game ad creatives — GameDev Reports · Dec 2025
- Ruling on PLR Worldwide Sales Ltd (Playrix) — Advertising Standards Authority · Sep 2020
- UK regulator bans misleading Homescapes, Gardenscapes pin puzzle ads — Game Developer · Oct 2020
- Entertainment: mobile app games — Advertising Standards Authority · May 2025
- Mistplay's 2025 Mobile Gaming Loyalty Index — Mistplay (Newswire) · Mar 2025
- Attention science and paid social strategy — Amplified Intelligence · Aug 2025
- Hacking the attention economy: the 1.5-second formula — VCCP Media · May 2025
- System1-TikTok study: branding elements and creator-led content — Chief Marketer · Aug 2025
- Key creative performance metrics — Motion · 2026
- Hook rate benchmarks — Billo · May 2026
- Demystifying creative diversification — Meta for Business · Dec 2025
