Noise
A platform where thousands of posters publish app promo slideshows and UGC on TikTok, paid per view.
Noise spread one slideshow format across thousands of paid-per-view posters, logging over 2.5 billion views in a year and about 30 cent installs, plus 54 million views in 30 days for its own app, which reached number 5 and 7 in its category. Tiny hook wording changes swung views from 10,000 to a million, and flooded niches faded. Lesson: test hooks constantly and rotate playbooks that don't pop within weeks.
- 1Brand sets the assetsScript, hooks, captions; CPM from install cost
- 2System randomizes combinationsEvery post unique, text pasted in TikTok
- 3About 50,000 posters publishAbout 100,000 slightly varied posts a day
- 4Pause what doesn't popCut playbooks within one to three weeks
- 5Giant hits plus long tailOver 2.5 billion views in a year
From Noise's own platform figures
What was said about Noise
1Noise's volume engine: thousands of posters paid per view
Spread one format across thousands of posters, not a few creators. Instead of paying 5 to 10 creators to post twice a day, Noise gives a pool of about 50,000 posters the exact assets, script and description, producing about 100,000 slightly varied posts a day. Founders on the platform logged over 2.5 billion views in a year.
The total came from a few giant hits (one new video reached 22 million views in 4 days) plus hundreds of thousands of posts earning 500 to 2,000 views each. Nobody planned the number; the team added it up at year end. The view and install figures come only from Noise's own platform and app, with no retention or revenue data.
Run viral UGC for reach and mass slideshows for conversion together. High-effort UGC with a soft call to action chases millions of views, while thousands of basic slideshows with a hard call to action reach the same warmed-up viewers. Noise's UGC hits got 3.8M, 5M and 2M views; the slideshows added only about 2 million views in 30 days but drove bottom-of-funnel downloads.
"These function very similar to retargeting campaigns."
Grid of 54,328 viral UGC posts, led by 5.09M and 3.8M views, about mandatory Secret Santa costs. Pay creators per view, with CPM set from your target install cost. No retainers: creators earn on CPM, and the brand sets the CPM from the cost per install it will pay, then adjusts levers like views per post. Noise averages about 30 cent installs on its own platform.
Creators get every asset and format and only manage an account and a posting schedule; they are not meant to invent content. Shotgun posters publish slideshows, while upskilled users make traditional UGC.
Viral 'why tf is Spanish so hard' TikTok crying-to-the-app format with 8.4M likes and 1.7M shares. Randomize slides and captions, and have posters paste text inside TikTok. Brands set captions, hook images, body images, descriptions and hashtags, and the system randomizes combinations so every post is unique. Posters paste the text in TikTok rather than baking it into the image.
Weber's theory, not a tested fact, is that native text signals a real user made the post. One campaign produced a huge number of variations across the 54 million views Noise saw in 30 days on TikTok, Instagram and YouTube.
Nick's TikTok account bookish.allison, where early book slideshows hit 887.7K, 775.1K and 257.7K views. Pause playbooks that don't pop within one to three weeks. Campaigns are run like Meta ad sets, with playbooks and posts underneath. One playbook got 446 posts in a week and only about 200,000 views, so it was paused, while 'how are we making money by Christmas' reached 9 million views.
The step before pausing is tweaking the prompts; if hundreds of posts still get little traction, a new idea takes the slot.
Low-view posts still convert because TikTok finds the right small audience. Videos with 200 to 500 views still drew comments asking what the app is and produced installs. Volume of small posts won't make you number one in a category, but it does produce downloads.
Weber also believes TikTok budgets views per niche each day: book slideshows got 1 to 2 million views each until thousands of accounts copied them, after which views spread thinner. This is his theory, so rotate formats as a trend ages.
Prove it manually before building the system. In July 2024 Noise paid 30 of its most engaged users to post Canva slideshows with an explicit call to action in a Brawl Stars niche for its Playbite app. The first slideshow did 20,000 views in a week, so they invited other founders in.
The founders came from performance marketing and chose TikTok because Meta ads cost a lot and underdelivered. "we like to do things that don't scale and then when we prove that they work, we scale them up"
SpyTok's Explore search for fitness, gym, workouts, showing outlier videos at 10.4x, 4.2x, 6.8x and 8.6x.
2How Noise grew its own money-making app
- Tells viewers what to do
- About 10,000 views
- Same post, new opening line
- About a million views
- 8 cent installs for about 3 days
New hook took a post from 10,000 to a million views
From Noise's own app results
'Don't get a second job' went viral where 'you need a side hustle' flopped. The post was modeled on a woman's Airbnb-cleaning-after-church video. Opening with 'Don't get a second job' made it take off, moving a post from about 10,000 to a million views, and the playbook ran at 8 cent installs for about 3 days.
Nobody likes being told what to do. Of the visuals tried, only washing dishes worked; vacuuming and pouring coffee did not. Four to six similar posts passed a million views.
Frame income as $20 a night, then $500 a month. Big claims like $7,000 in a week get dismissed as a scam even with receipts or bank stubs. 'I make noise posts every night for $20' got several videos over a million views, and commenters asked how it works instead of calling it a scam.
The monthly total anchors on something real, like a $498 car payment. "Anyone can imagine a $20 bill in their hand, but then they're bad at math."
Never name the app in the video; let seeded comments sell it. The app appears only in the description, and many posters comment that it worked for them. One comment, 'My work has this too. I use the noise app to make extra money', got about 1,686 likes, which Weber estimates means at least a thousand downloads or searches.
A secret Santa post complaining about a $60 mandatory gift was built to draw comments. Seeding needs many accounts, so it is hard to do alone. These tactics took the app to number 5 and number 7 in its category and 54 million views in 30 days.
3Copying formats that already went viral
Slot your app into a proven trend instead of inventing formats. Lingo Pingo's format, a person crying over a word while the app roasts them, was copied from a trend a random woman started with Google Translate and got about 20 million views overnight. It ran through October and November in about 15 variations.
The idea mirrors the 'roast me' subreddit, done by an AI. Creators got an example, exact prompts and an explanation of why it is funny, and it works in any language. "We don't have to be super creative. We just have to have a good eye for what's already gone viral"
Find winning formats by studying outlier videos, then test small. Noise's SpyTok scans hundreds of thousands of app-promoting TikToks to surface the most viral; a fitness-app search turned up a 106x outlier using a text hook over an image and a call to action on slide two.
Formats are tested with a small group of posters before scaling. SpyTok is used by 11 Labs, Ammo and Voodoo. Reddit helps to learn how an unfamiliar niche talks, but the best inspiration is the platform you post on.
4Hooks and captions that swing views 100x
Use hook images viewers understand and relate to instantly. In clean-eating slideshows, hook images with hard-boiled or scrambled eggs consistently got higher views, first for one brand and then across many accounts.
Eggs are healthy, cheap and understood at a glance, and egg prices had been an election issue. A relatable hook makes viewers swipe to the next slide, which signals TikTok to push the post.
Use AI for caption variations but edit them by hand. AI-generated captions don't yet feel organic enough, so ChatGPT is used for variations that are then curated. AI-prompted captions have still hit 1 to 2 million views.
"the ones that you spend the least amount of time on for some reason are the ones that always go viral."
Meet The Guy Who Solved App Growth on TikTok
Founders on the platform logged over 2.5 billion views in a year, with about 30 cent installs on its own platform. It also promoted its own money-making app with 'Don't get a second job' posts and seeded comments, reaching number 5 and number 7 in its category, and did 54 million views in 30 days.
Spread one format across thousands of posters, not a few creators. Instead of paying 5 to 10 creators to post twice a day, Noise gives a pool of about 50,000 posters the exact assets, script and description, producing about 100,000 slightly varied posts a day. Founders on the platform logged over 2.5 billion views in a year.
The total came from a few giant hits (one new video reached 22 million views in 4 days) plus hundreds of thousands of posts earning 500 to 2,000 views each. Nobody planned the number; the team added it up at year end. The view and install figures come only from Noise's own platform and app, with no retention or revenue data.
Run viral UGC for reach and mass slideshows for conversion together. High-effort UGC with a soft call to action chases millions of views, while thousands of basic slideshows with a hard call to action reach the same warmed-up viewers. Noise's UGC hits got 3.8M, 5M and 2M views; the slideshows added only about 2 million views in 30 days but drove bottom-of-funnel downloads.
"These function very similar to retargeting campaigns."
Grid of 54,328 viral UGC posts, led by 5.09M and 3.8M views, about mandatory Secret Santa costs. Pay creators per view, with CPM set from your target install cost. No retainers: creators earn on CPM, and the brand sets the CPM from the cost per install it will pay, then adjusts levers like views per post. Noise averages about 30 cent installs on its own platform.
Creators get every asset and format and only manage an account and a posting schedule; they are not meant to invent content. Shotgun posters publish slideshows, while upskilled users make traditional UGC.
Viral 'why tf is Spanish so hard' TikTok crying-to-the-app format with 8.4M likes and 1.7M shares. Randomize slides and captions, and have posters paste text inside TikTok. Brands set captions, hook images, body images, descriptions and hashtags, and the system randomizes combinations so every post is unique. Posters paste the text in TikTok rather than baking it into the image.
Weber's theory, not a tested fact, is that native text signals a real user made the post. One campaign produced a huge number of variations across the 54 million views Noise saw in 30 days on TikTok, Instagram and YouTube.
Nick's TikTok account bookish.allison, where early book slideshows hit 887.7K, 775.1K and 257.7K views. Pause playbooks that don't pop within one to three weeks. Campaigns are run like Meta ad sets, with playbooks and posts underneath. One playbook got 446 posts in a week and only about 200,000 views, so it was paused, while 'how are we making money by Christmas' reached 9 million views.
The step before pausing is tweaking the prompts; if hundreds of posts still get little traction, a new idea takes the slot.
Low-view posts still convert because TikTok finds the right small audience. Videos with 200 to 500 views still drew comments asking what the app is and produced installs. Volume of small posts won't make you number one in a category, but it does produce downloads.
Weber also believes TikTok budgets views per niche each day: book slideshows got 1 to 2 million views each until thousands of accounts copied them, after which views spread thinner. This is his theory, so rotate formats as a trend ages.
Prove it manually before building the system. In July 2024 Noise paid 30 of its most engaged users to post Canva slideshows with an explicit call to action in a Brawl Stars niche for its Playbite app. The first slideshow did 20,000 views in a week, so they invited other founders in.
The founders came from performance marketing and chose TikTok because Meta ads cost a lot and underdelivered. "we like to do things that don't scale and then when we prove that they work, we scale them up"
SpyTok's Explore search for fitness, gym, workouts, showing outlier videos at 10.4x, 4.2x, 6.8x and 8.6x. 'Don't get a second job' went viral where 'you need a side hustle' flopped. The post was modeled on a woman's Airbnb-cleaning-after-church video. Opening with 'Don't get a second job' made it take off, moving a post from about 10,000 to a million views, and the playbook ran at 8 cent installs for about 3 days.
Nobody likes being told what to do. Of the visuals tried, only washing dishes worked; vacuuming and pouring coffee did not. Four to six similar posts passed a million views.
Frame income as $20 a night, then $500 a month. Big claims like $7,000 in a week get dismissed as a scam even with receipts or bank stubs. 'I make noise posts every night for $20' got several videos over a million views, and commenters asked how it works instead of calling it a scam.
The monthly total anchors on something real, like a $498 car payment. "Anyone can imagine a $20 bill in their hand, but then they're bad at math."
Never name the app in the video; let seeded comments sell it. The app appears only in the description, and many posters comment that it worked for them. One comment, 'My work has this too. I use the noise app to make extra money', got about 1,686 likes, which Weber estimates means at least a thousand downloads or searches.
A secret Santa post complaining about a $60 mandatory gift was built to draw comments. Seeding needs many accounts, so it is hard to do alone. These tactics took the app to number 5 and number 7 in its category and 54 million views in 30 days.
Slot your app into a proven trend instead of inventing formats. Lingo Pingo's format, a person crying over a word while the app roasts them, was copied from a trend a random woman started with Google Translate and got about 20 million views overnight. It ran through October and November in about 15 variations.
The idea mirrors the 'roast me' subreddit, done by an AI. Creators got an example, exact prompts and an explanation of why it is funny, and it works in any language. "We don't have to be super creative. We just have to have a good eye for what's already gone viral"
Find winning formats by studying outlier videos, then test small. Noise's SpyTok scans hundreds of thousands of app-promoting TikToks to surface the most viral; a fitness-app search turned up a 106x outlier using a text hook over an image and a call to action on slide two.
Formats are tested with a small group of posters before scaling. SpyTok is used by 11 Labs, Ammo and Voodoo. Reddit helps to learn how an unfamiliar niche talks, but the best inspiration is the platform you post on.
Use hook images viewers understand and relate to instantly. In clean-eating slideshows, hook images with hard-boiled or scrambled eggs consistently got higher views, first for one brand and then across many accounts.
Eggs are healthy, cheap and understood at a glance, and egg prices had been an election issue. A relatable hook makes viewers swipe to the next slide, which signals TikTok to push the post.
Use AI for caption variations but edit them by hand. AI-generated captions don't yet feel organic enough, so ChatGPT is used for variations that are then curated. AI-prompted captions have still hit 1 to 2 million views.
"the ones that you spend the least amount of time on for some reason are the ones that always go viral."
More in Marketing & growth tools
Marketing & growth toolsApp Masters
An app marketing and growth consultancy run by Steve P. Young, which also hosts a podcast.
115 sources · 465 tipsApple Search Ads
Apple's ad platform for promoting apps in the App Store.
44 sources · 132 tips
Marketing & growth toolsMeta
The parent company of Facebook, Instagram, WhatsApp and Threads, run by Mark Zuckerberg.
42 sources · 106 tips
Marketing & growth toolsSensor Tower
An app intelligence and ASO platform with keyword rankings, competitor research and download estimates.
42 sources · 90 tips
Marketing & growth toolsMobile Action
An app store optimization and mobile marketing intelligence platform.
28 sources · 81 tips
Marketing & growth toolsHubSpot
A CRM and marketing software platform built for small and medium businesses.
14 sources · 41 tips
Marketing & growth toolsAdMob
Google's mobile ad monetization platform with mediation, open bidding and unified payments for app developers.
10 sources · 47 tips
Marketing & growth toolsAppsFlyer
A mobile attribution and marketing analytics platform for app marketers.
17 sources · 30 tips