UGC Handbook / Lesson 01 of 14 / Chapter I: Before you post

41 million views, zero revenue: what UGC actually does for subscription apps

One video. And the company behind it shut the product down.

The video came from the founders of Coconote, an AI note-taking app that rode UGC to $6.7M ARR and a Quizlet acquisition in two years. They know what they're doing. That 41-million-view hit still produced almost nothing, and the reason why is the best possible starting point for this handbook.

The 41-million-view funeral

Coconote's team ran a side experiment called pdftobrainrot.com: upload a long PDF, get it read back to you over Minecraft parkour footage. True brainrot.

The internet loved it. 41M views and 4.5M likes on a single video.

Then almost nobody paid. Co-founder Zack Hargett named the mistake on the Sub Club podcast: "framing your product as a novel toy and not a solution to a problem." A toy gets attention. It doesn't get subscriptions. They sunset the product.

Here's what makes this useful instead of just painful. The same founder had seen the opposite happen before: at his previous startup, the first video ever posted on a fresh TikTok account hit 8 million views and drove 150,000 signups in two days (to a web product; Zack reckons it would have topped the App Store had it been an app).

Same person. Same platform. Both viral.

Zack's own takeaway is that great content wins. But put the two videos side by side and one difference is hard to miss: one framed the product as the fix for a real problem, the other framed it as entertainment.

So before you script anything, complete this sentence:

Every video you make either answers that in the first few seconds, or it's content made for the algorithm, not your business.

It's a portfolio, not a lottery ticket

Individual videos are nearly impossible to predict. Experienced UGC operators don't try.

Ajay Mehta, whose AI companion app Tolan grew on creator content, put the hit rate plainly: they fix a set of angles (cooking with your Tolan, doing makeup, relationship talk), run many variations inside each angle, and "one out of maybe 20 videos kind of pops off." When it does, it can reach escape velocity. One cooking video did 10M views.

The other 19 are the cost of finding it.

Paid UGC runs on the same math, at sharper odds. One team interviewed on Superwall's channel found that just 1 in 85 ad creatives became a true winner, and of the $1.5M to $1.8M they spent, roughly $800K was, in effect, the search cost of finding those winners.

This changes what "good at UGC" means. It's not writing one brilliant script. It's:

• committing to a batch (20+ variations) before judging anything
• keeping each unit cheap enough that 19 duds don't hurt
• judging the portfolio, not the video

If you plan for one video to work, every flop feels like failure and you'll quit in week two. If you plan for 5% of videos to work, flops are just the meter running.

Annoying, but true.

Copy winners. Don't invent them.

Your niche has already run the experiment for you.

The 80% rule. A founder whose app makes $2M a year put a number on it in a Superwall interview: "80% has to be copying what's already gone viral, because that's what's going to go viral again." Copying here means structural copying: same hook mechanics, same pacing, same shot types, with only the product and the words swapped. The other 20% is for genuinely new experiments.

How literal should the copying be? More literal than feels comfortable. When the app Mojo expanded to Brazil, they first tried adapting their winning ad with local creators. It underperformed.

So they replicated it instead: AI-dubbed the original video, preserving the exact timing, pauses, gestures, and energy. Result: acquisition cost 40% lower than native creators. Their conclusion: "execution mechanics matter just as much as the message."

The market is funding the wrong thing, provably. AppsFlyer analyzed 1.1 million ad creatives across $2.4B in spend. For social apps, tutorial and app-review formats deliver 45% higher IPM (installs per thousand impressions) and 17% better day-7 retention than testimonials. Yet testimonials still capture the majority of budgets.

Most teams fund what's familiar, not what performs. Copying actual winners is a real edge.

Do this before you script anything

1. Write the problem sentence. Fill in the blank above. Toy test: would people share this video without wanting the product? If yes, reframe.
2. Collect 5 winning videos in or near your niche. Search your problem keywords on TikTok, sort by views, save the top performers from the last 90 days.
3. Pick ONE format to copy. Write down its fixed mechanics: hook structure, length, audio type, shot sequence.
4. List 20 variations you'll make inside that format: different opening lines, angles, situations. Only the content varies. The mechanics don't.
5. Decide your signal now, before posting: engagement rate ((likes + comments + saves + shares) ÷ views) of 7.5%+ is a real signal. Raw views alone are not. Remember the funeral above.