Most apps that try short-form video start by copying the last video they saw with five million views. The copy usually stalls at a few hundred, because a big number on its own tells you almost nothing about why a video spread. Apps that get views month after month look for formats instead: structures that keep winning on many different accounts. Then they rebuild those structures around their own product.
The method below works the same on TikTok, Instagram Reels and YouTube Shorts. It runs from the first search to the week a format starts to fade.
What is a viral video format?
A video format is the reusable structure of a short video, beat by beat, from the first frame to the payoff. You can swap the topic and rewrite the script, and the format still holds.
The common production styles in app marketing are talking head, green screen, phone-in-hand demo, slideshow or carousel, storytime over filler images, text over driving or Minecraft footage, faceless storytelling, AI avatars over slides, and AI voiceover on b-roll. A format sits one level below the style: it’s the specific sequence of beats inside it.
Every video has three layers worth pulling apart. The format is the sequence. The angle is the promise the video makes. The hook is whatever stops the scroll in the first second or two.
Cardstock, a trading-card scanning app, built most of its videos on one sequence: a rare card on screen for three to five seconds, the scan, then the price. Its founder says his two card apps bring in about $41,000 a month together. The angle is “what is this card worth?”, and the hook is the card itself, held close to the camera before anyone speaks. A plant app or a wine app could run the same sequence with its own angle and hook. That’s why one good format can carry dozens of videos.
Start with what your app shows on camera
Apps are harder to film than physical products. Before any research, write down the one moment where your app visibly does something. A calorie app photographs a plate and the numbers appear. A study app turns a stack of lecture notes into a quiz. Time that moment. If the payoff takes more than a few seconds, or only makes sense with a voiceover, most of the formats you find won’t have room for it.
If your app has no such moment yet, build one first. Ship a TestFlight build with only the most visual feature, and give it to a creator to film. That costs far less than months of content about a product nobody can see working. Narrow features also film better. “A personal finance assistant” is hard to show, while a scan that reveals what you spent on coffee this month fits in a single shot.
Find the words your audience uses
Search only finds what you type, so your search phrases should come from the people you want to reach. Good sources are Reddit threads in your niche, app-store reviews (your competitors’ as well as your own), the comments under popular videos, and your competitors’ ad libraries. Copy the phrases people repeat, word for word.
In comment sections, two kinds of comment deserve a note in your file: “what app is this?” and “how do I…?” Both mark a video that made viewers want a solution. Also watch for comment sections where viewers argue or swap advice with each other, since bought views almost never come with that.
Scroll every day, and save by format
Fifteen minutes a day across TikTok’s For You feed, Reels and Shorts finds more formats than an occasional two-hour binge, because formats rise and die within weeks. Use a separate research account on each platform. Follow and interact with creators in your niche until each feed mostly shows your audience’s content. Then spend part of the session outside the niche, where you’ll find structures nobody in your category has used yet.
Save what you find into folders named after the format, such as “object, scan, reveal”, “before-and-after slideshow” or “day in the life”. A skincare slideshow from Reels and a budgeting slideshow from TikTok belong in the same folder, because you’re collecting structures. Write down each video’s post date and the date you saved it. A video still climbing a week later is a stronger lead than one that peaked on day one.
On TikTok, the Creative Center is worth a weekly visit. It filters trending videos by industry, region and date range, which makes it a good second source next to your feed. It only covers TikTok, so on Reels and Shorts, rely on each app’s own search.
Search outside your niche
Your own niche mostly shows you what competitors already post. Run every search phrase on TikTok, Instagram Reels and YouTube Shorts, at three distances:
- your niche and your competitors’ accounts
- nearby audiences who share the problem, or the same audience with other interests
- unrelated niches whose structure could hold your product
Some structures travel almost anywhere. Listicle slideshows and short background clips with text on top work in nearly every niche, and on all three platforms.
Two sources get skipped a lot. The first is creators’ sponsored posts, which show formats brands already chose to pay for. Brands often buy posts in bundles, though, so seeing one brand five times on a creator’s page doesn’t prove the first post did well. The second is ordinary people posting about the problem your app solves, which is especially useful when your category has few competitors. Their best videos are often formats waiting for a product.
How do you spot an outlier video?
Judge every video against the account that posted it, on the platform where it was posted. Take the median views of the account’s recent posts and divide the video’s views by that number. The result is the video’s multiple.
A 60,000-view Reel on an account that usually gets 2,000 has a multiple of 30. A 2-million-view TikTok on an account that always gets 2 million has a multiple of 1, and it teaches you nothing.
As a rule of thumb, a video at 5 to 10 times its account’s usual views is a real outlier. Small accounts are the best place to look: an account under 5,000 followers with a video past 100,000 views has almost certainly hit on a structure. When you compare a batch of similar videos, the formats are usually in the top 10% by multiple.
Use the median (the middle value) and skip the average. One old hit can pull an account’s average so high that its next outlier looks ordinary.
Check that it works on many accounts
A single viral video can owe everything to the person in it. A creator’s personal delivery, timing or luck can carry one post, and remakes of that post by other creators often fall flat.
Before you call a structure a format, find it working on accounts that have nothing to do with each other. Five to twenty separate posts and accounts is a good bar. Then ask whether it can carry ten or more videos that are meaningfully different from each other. If it only works once, it’s a joke, and a joke doesn’t make a format.
Expect big swings even inside a format that works. Massive, an AI app that applies to jobs for you, makes about $200,000 a month. Its founder says two nearly identical videos in the same format got 1.7 million views and 382 views. Judge a format on a batch of videos.
Remove the fake signals
Some views and comments say nothing about the structure. Clear them out of your file before you decide anything.
Paid boosts are the most common. Promotion adds views without the likes and comments that real reach brings. A rough check: 50,000 likes on 2 million views reads as organic, while 2,000 likes on 2 million views usually means paid reach.
Planted comments come next. Some app creators pin their own “What’s this app?” comment under their videos, and comment seeding is openly taught as a tactic. A wall of curious comments can be manufactured.
Clickbait is easy to catch once you look at the likes. A video with hundreds of thousands of views and only a couple of hundred likes broke its promise: people watched it and left disappointed.
Moments need care. A video riding a news story or a celebrity may have spread because of the moment itself. Borrowed celebrity audio also carries legal risk, and it has drawn cease-and-desist letters. Count a moment-driven video only if its structure works again with a different moment.
Ad galleries come last. TikTok’s Top Ads and Meta’s Ad Library show what brands paid to run. They’re useful for ideas and tell you nothing about what spreads on its own.
Break the video down before you adapt it
For each outlier that survives, write the video out beat by beat with rough timestamps. Note the first frame and the first line, what happens at each point, when the product or subject appears, the text on screen, how the sound is used, and the call to action. Then cross out the topic and read your notes back. If they still make sense with any subject dropped in, you’ve written down the format. Put your app where the original subject was.
Writing it out catches things you miss while watching. A seven-second driving clip with text over it gets replayed, for example, because the text takes more than one viewing to finish reading. That loop works in any niche and on any of the three platforms.
Change about 10 to 20 percent of a reference, and make the script and footage your own. Be careful about pairing the hook from one video with the demo from another. The original may have worked because of how its opening led into its payoff.
Test a few formats, one change at a time
Start with about three formats, each a truly different structure. A dozen rewrites of one script only tests one idea.
Inside each format, keep the structure and the product moment fixed and change one element per round: the first image, the opening line, the caption or the pacing. Run the same demo with different hooks across many creators, even dozens, and you’ll learn which opening carries the format. Cardstock’s founder says they kept the scan-and-reveal sequence fixed and tested only the hook text, the first image and the pacing.
Post every test on TikTok, Reels and Shorts, because the same video can perform very differently from one platform to the next. Read each platform’s results on their own, against each account’s usual views there.
When a video takes off, make more of it quickly. Once a post passes 10,000 to 20,000 views with at least 3% engagement, repost it, change the text hook, or reshoot the same script in the same setting. Expect remakes to fall short of the original: the typical remake of a viral post gets about a fifth of the original’s views, while a few do two to six times better. Making five to ten versions of each winner tends to win back the original’s total views, and one of them may beat it.
How long does a viral format last?
Most formats last a few months, usually three to six, before views drop. Pingo, an AI app for practicing a new language out loud, makes about $500,000 a month, according to its founder. He says its main format lasted four to five months.
To see a format fading, compare its newest videos with its earlier ones, each measured against the account’s usual views on that platform. When the new ones keep landing lower, change the angle or the hook, and start looking for the format that will replace it.
The broad styles wear out as well. In app marketing, influencer deals saturated in 2021, mass UGC in 2022, slideshows in 2023, AI avatars over slides in 2024 and fully generated avatars in 2025. Whatever replaces the current wave is usually already in someone’s feed.
Frequently asked questions
How do you find viral video formats for your app on TikTok, Reels and Shorts?
Search all three platforms with the words your audience uses, both in your niche and in unrelated niches. Compare each video’s views with its account’s median to find the outliers. Then check that the same structure works on several separate accounts before you rebuild it around your app.
What is an outlier video?
An outlier is a video with far more views than its account usually gets. You measure it by dividing the video’s views by the median views of the account’s recent posts. At 5 to 10 times, the structure did something the account normally can’t.
How many views does a short video need to be viral?
No single number works across accounts or platforms. For a new app account, 5,000 views is a reasonable first checkpoint. On an account under 5,000 followers, a video past 100,000 views is a strong signal. The multiple against the account’s usual views is the fairer test.
Can you copy viral videos from other apps?
Copy the structure and change the rest. The typical remake of a viral post gets about a fifth of the original’s views, and copying someone’s audio or face can get your video taken down.
How many video formats should an app test at once?
Start with about three formats that are truly different from each other. Inside each one, change one element at a time, and post every test on TikTok, Reels and Shorts.
How long do short-form video formats last?
Usually three to six months before views fall off. Compare a format’s newest videos with its earlier ones on each platform, and refresh the hook or angle when the new ones start landing lower.