Pingo AI · B2C · AI language speaking app
Scale winning formats through a pay-per-view creator network; test ideas with core creators
Winning variations get scaled quickly; spend is easy to flex; no numbers
What they did
What happened
In their words
Morrie Schonfeld - How I built a $500k/mo ai app (so you can just copy me)
Morrie Schonfeld… value their time, but really like the money is in the bonuses. So like the numbers are like 50,000, 100,000, 200,000, 500,000, and a million views. So we have these set milestones. They're not stacking. So if a creator gets $100 at 100,000 views and $50 at 50,000 views, they don't get 150 when they get to 100,000 views, they get 100. So it's kind of like you earn the highest tier. They don't stack, if that kind of makes sense, plus the base pay.
Morrie SchonfeldAnd then that way, creators are motivated to make the content cuz they're getting paid for their time, and motivated to go viral. And then on something like Noise that you mentioned, like we talked about this earlier, so we we use Noise for a lot of distribution to scale what really works, and also really quickly test ideas, that's on a CPM model with a cap. So we kind of have these two different spend buckets. We have our in-house program with base pay, fixed tier bonuses that we manage, and we have Noise with their creators, which is CPM post cap.
Morrie SchonfeldAnd it makes spending interesting cuz we can really easily change our Noise budget, whereas our in-house spend is a lot more fixed because of the way we do it and our creator base, and we can't just like turn it off. And then we use Noise for kind of like scaling what works and really quickly testing, and then with our creators, we have a cohort of creators that we use to test ideas that we know are really good, we know how they make content, …
Morrie Schonfeld… we've gotten really deep into paywall testing recently, and I've gotten granular. But when it comes to content, we'll give them like a brief. We'll say, "All right, here's like an example video." If it's something one creator's made before, like we had a video a format that send recently, like sort of like a modulation of like the mean crying Pingo, but a bit different, one creator did it and went really viral, and we took that and scaled it.
Morrie SchonfeldAnd so we said, "All right, here's an example, and here's why we think it went viral," kind of like what Noise likes to do. Like, "Here's why it went viral. Here's the key steps to recreate, but recreate this as close as possible in your target language. Like go." That's kind of what we do both in-house and on Noise. And if it's a format that we haven't ever made before, right, we think of like, "Okay, this is a really cool idea that no one's done it," then we'll make an example. Like, "All right, here's like an example. Here's what we're looking for."
Joseph ChoiThat is unique. I think there's not that many founders who are willing to get that in the weeds and like literally make the videos before their own creators. I've talked to like successful founders who do UGC, and a lot of them actually do say this. It does seem like this is the way to do it. Like are most of the ideas sourced from you, or do your creators come up with ideas and then you actually execute them, and then if it goes well, then you …
Get tactics like this every Monday
New tactics from the week's founder interviews, each linked to where it was said.
More from this episode
- Only build an app whose main feature looks interesting on videoTalking to the AI became the core of every video; app reached ~$500K/month
- Film yourselves talking to the app on TikTok, one account per languageReached $20K MRR in about 3 months with almost no outside creators
- Show people using the app in every video, not just talk about the topicLanguage-topic videos without the app flopped; app-demo videos went viral and converted
- Tie videos to beginner lessons and culture moments, like K-dramas or the TikTok banA 'basic Korean quiz' got 77K views; a TikTok-ban/RedNote video got 1M
- Film the AI tutor roasting a learner until they cry on cameraTop video hit 52M views; the format has worked in every language since September
- Let users write their own scenes for the AI so creators can script funny videosCreators found the viral 'mean Pingo' format with it; users now ask for a mean mode
- Post language-tip slideshows that end with a pitch for the appSome hit 1M views, but accounts burned out fast; weak-CTA slides converted far worse
- Have assistants DM small creators learning the language, then move them to emailBuilt a roster of ~100 creators across ~20 languages
- Pay creators a base per video plus bonuses at 50K to 1M viewsKeeps creators paid for their time and chasing viral hits; no numbers
- Treat creators like investments: drop the weakest, keep the stars, train the middleTop 20-30% of ~100 creators drive most views; cut 30-50 who drove none
- Coach creators on lighting, camera angle, sound and doing something on cameraJust started; small tweaks tend to help good creators, no numbers yet
- Make viral videos for learners of RussianAbout half of new users came from Russia and couldn't pay; Russian content killed
- Show a timeline on the paywall: free today, reminder before billing, cancel anytimeBeat social-proof and yearly-vs-monthly paywall versions; no numbers given
- Offer the free trial only on the yearly plan and hide the monthly one
- Test a 3-day against a 7-day free trial, US and abroad separatelyNear-even in the US; 7-day converted ~2% better internationally (early data)
- When someone backs out of a 3-day trial, offer 7 free days at a lower priceConverts better than showing the 7-day trial first; exact lift not given
- Bill US users through web checkout to keep more of each sale
- Join Y Combinator, then raise quickly on strong revenue and CAC/LTV~100 investor meetings booked; round closed in about 3 days
Tags: tiktok, cpm-creators, format-testing, creative-brief, core-creators