Savannah Sanchez's team makes hundreds of ads a week for ecommerce brands. In a webinar with AppLovin she explained why the ads that win there break most paid social rules. Here's what she said, and what it means for brands, apps, and games.
81%Of the ads AppLovin serves run longer than 30 seconds
50%+Of clicks in a creative set come from the end card
8Ads a buyer sees on average before they purchase
The flipped formula
A social ad peaks in second one. An AppLovin ad peaks at the end.
On a feed, the first three seconds decide everything. Inside a game, the viewer is already watching and counting down to their next level, so the story can build. Her pacing for a 60-second AppLovin ad starts low, rises from about 20 seconds, peaks, and closes on the offer.
Meta and TikTokStop the scroll, then hold on
HookAttention falls away as the feed pulls
AppLovinSet the scene, build, then sell
Set the stageRising actionClimaxCTA
0s20s40s60s
The shape of the arc she described, not measured data. Her example: a creator cooks up a hair serum, pulls it from the oven, and the ad ends on the website and offer.
The viewer
Forget the scroll.
AppLovin ads run full screen inside games like Subway Surfers, Candy Crush, and Words with Friends. Nobody can swipe past them, so the first three lessons undo a few paid social habits.
01
Treat it like a TV spot
On rewarded placements a player chooses to watch in exchange for coins or a new life. That changes the mood from I'm annoyed by ads to I'm choosing to watch this. Every ad is unskippable for at least five seconds, and full-screen ads only appear every 2 to 5 minutes of play.
The fixUse the time to educate, build trust, handle objections, and sell.
02
Run 30 to 60 seconds, not 15
Meta and TikTok push ads toward 15 to 30 seconds. On AppLovin, 81% of served ads run past 30 seconds, and her best performers lean to 45 to 60.
The fixCut your AppLovin versions long. A trimmed Meta ad wastes the minute you've been given.
03
Save your best for the end
Attention peaks in the last 5 to 10 seconds, because the player is almost back to their level. The hook still sets the tone, but it doesn't have to carry the product, the offer, and the proof.
The fixPut the product in use, the offer, and the CTA in the back half of the ad.
The ad itself
Tell a story, then sell.
Her top AppLovin ads looked nothing like her top Meta ads. The next three lessons are about what goes on screen, and the screen after it.
04
Story beats flashy editing
The winners she pulled had a voiceover, easy captions, a creator talking to camera, and a reason to buy. Demos, unboxings, testimonials, and price-justifying maths worked. Studio footage didn't make the list.
The fixScript an arc with a resolution, then do the hard sell in the last five seconds.
05
Borrow from gaming advertisers
Gaming advertisers are the top spenders on AppLovin and set most of its creative norms. Playful, game-like edits translate, like a creator tapping the product out of her phone. Jarring scroll-stoppers feel out of place inside a game.
The fixStudy gaming ads, and the longer educational ads brands run for retargeting on Meta.
06
Test end cards like videos
The interactive screen after the video drives more than half the clicks in a creative set, and it opens the same landing page as the video. When a video starts winning, scale it by giving it more end cards.
The fixPair one proven piece with one new one: a new video with a proven end card, or the reverse.
End card formats she'd test first on a winning video
“Good advertisers use their existing social ads. Great ones adapt.”
Savannah Sanchez, The Social Savannah
Gaming teams worked this out first, and that's where we started. The video earns the attention, the end card takes most of the clicks, and the page closes the sale. That's why we build the ad, the end card, and the landing page as one connected test, then improve it from purchase data.
Gaming companies upload tens of thousands of ads to AppLovin. Even top ecommerce brands usually test hundreds. The last four lessons close that gap.
07Upload everything you've already madeStart with every ad you've run on Meta or TikTok, including the ones that flopped. Her top AppLovin performers were rarely the Meta winners, and the algorithm sorts the pile quickly.The fixLoad the back catalogue on day one, then build long-form originals while it learns.
08Make volume and variety a habitTest thousands of ads, not hundreds. Variety means different creators, messages, and styles, not one ad with a hundred hooks. For a new account she suggests 4 to 12 new concepts a week. Her team shoots Monday and Tuesday, edits Wednesday and Thursday, and clients launch on Friday.The fixStitch existing creator footage into 60-second mashups. They count as new ads without a new shoot.
09Don't kill the big spender too soonIn one client account, buyers saw about eight ads before purchasing. The first ad they saw had high spend and the lowest ROAS in the account, but it started most of those journeys.The fixTreat AppLovin's spend as the first signal. Give new ads a few weeks, in one-month sprints, before you call them.
10Change one thing at a timeWhen an ad wins, work out why. Is it the creator, the script, the hook, or the format? Keep the rest exactly the same and swap one.The fixIterate off your best ad one variable a week: same script with a new creator, or a new hook on the same body.
Want ads that hold the full minute?
Book a call and name three rivals. We'll read your live ads against theirs: what you all say, who's on screen, and what's lasting.