The Superwall Podcast · 2025-08-04
Yev Marusenko - How I Grow Viral Apps to 1M+ Downloads (just copy me)
Yev Marusenko, head of growth at AI learning app Sizzle AI and owner of 10+ faceless YouTube Shorts channels, on combining organic and paid: put a few dollars behind every organic creator video to surface outliers (20-30 videos per winner), scale winners with headline and country variants, and use the winning ad as a traffic firehose for weekly in-app paywall and onboarding tests (cost per purchase $500 to $300 in days). Also: fixing the biggest funnel drop-off for 2-3x jumps, attribution triangulation (reported purchases ~half of real), and a $5K test budget plan.
Play the episode12 tactics
Numbers mentioned
| Subject | Metric | Value | Quote |
|---|---|---|---|
| Sizzle AI | drop off Largest single funnel step drop-off the team focused on (over 50%) | 50% approximate | “as soon as we shifted focus knowing on there's over 50% drop-off here, when we improve that, we see these 2X to 3X growth moments” |
| Sizzle AI | growth rate Users reaching the next funnel stage after fixing the biggest drop-off, vs ~10% gains from feature optimizations | +2–3 x approximate | “And all of a sudden, you just 2X, 3X how many people make it to the next stage.” |
| Sizzle AI | count Likes on the showcased creator TikTok (JoinBrands creator) later amplified as a paid ad | 79K likesper video | “This video got 79,000 likes.” |
| Sizzle AI | count Organic creator videos it took to find one clear winner | 20–30 videos approximate | “This actually took more like between 20 and 30 organic pieces of content to find one that's way better than the rest and then kind of replicating that.” |
| Sizzle AI | new customers Purchases attributed to the single winning TikTok creative run as a paid ad | hundreds of purchases approximate | “specifically for this video, it's into hundreds of purchases. So, it's not like crazy where it's like thousands of purchases from a particular video.” |
| Sizzle AI | cpa Lenient cost-per-purchase target for ads whose job is feeding data into onboarding/paywall tests | USD 200per purchase goal | “So, if we are able to reach $200 cost per purchase, which is way more lenient than if we're like being more aggressive on we need to reach profitability where some of these ads are like $100 cost per purchase.” |
| Sizzle AI | cpa Cost-per-purchase goals for profitability-focused ads (US targeting); $50 and $100 tiers | USD 50–100per purchase goal | “we have different goals on trying to reach $50 cost per purchase versus $100 cost per purchase versus $200 cost per purchase” |
| Sizzle AI | conversion rate Install -> purchase on traffic sent into the onboarding flow | 1% approximate | “I'm kind of like sending in traffic and knowing that we're getting 1% conversion rate from the install all the way down to the purchase.” |
| Sizzle AI | cpa Cost per purchase before rapid in-app testing (varies by targeting) | USD 500per purchase approximate | “right away that cost per purchase goes from $500 to $300 per, depending on what I'm targeting” |
| Sizzle AI | cpa Cost per purchase within days of 3-4-way paywall and in-app experiments on winning-ad traffic | USD 300per purchase approximate | “right away that cost per purchase goes from $500 to $300 per, depending on what I'm targeting” |
| Sizzle AI | churn Average subscriber lifetime used to estimate LTV | 3–7 months approximate | “we know what that our churn is depending on the audience, the country, it's kind of between three and seven months” |
| Sizzle AI | ltv Estimated value per purchase; higher for some products, countries and iOS | USD 30–70per purchase estimate | “it's $30 or like $70 or like higher kind of depending on different parts of the product that is purchased, depending on the country, and also device, like iOS is much better” |
| Sizzle AI | share Part of paid traffic held to profitability targets; the rest used for aggressive in-app testing | 20% approximate | “then I'm taking where like 20% of it is matching our numbers exactly, so it's kind of I'm okay with incremental improvement there while crazy testing is here” |
| Sizzle AI | count Headline variants tested on one winning creative | 10 headlinesper creative approximate | “Incremental meaning that then taking this creative and testing 10 different headlines.” |
| Sizzle AI | attribution gap Real purchases vs purchases reported in the ad platform (usually about double) | 2 x estimate | “if you run a ad and you have 10 purchases reported or 100 purchases, what does that actually mean? Are is the real data where it's like double that? So, I think there's like a good insight where it's usually like double that” |
| Sizzle AI | share At least this share of purchases arrive later than 1 day after the ad (1-day vs 7-day window) | 10% approximate | “knowing how many are coming in right away, kind of like 1 day versus 7 day or beyond, to know that at least 10% of the purchases are coming in later that aren't relatively instant” |
| Sizzle AI | share Extra purchases seen via SKAN and other attribution methods beyond platform-reported | 10–20% approximate | “Then also looking at SKAN versus uh what's other reported purchases, knowing that another over 10 to 20% of the purchases are coming in through these other attribution methods.” |
| Sizzle AI | share Additional spillover purchases (saw ad, later searched the App Store) seen in App Store Connect analytics | 10–20% approximate | “And then I'm actually in Apple Connect looking at analytics like based on platform, knowing that another 10 to 20% of purchases are coming in from spillover types of attribution that are happening.” |
| Sizzle AI | count Ads running across ad groups | thousands of ads approximate | “there's like I have like hundreds of ads, actually like thousands of ads.” |
| Sizzle AI | conversion rate Purchase conversion of users who start with a particular product/journey vs users roaming the app | +10 x approximate | “some of them start using a particular product and end up purchasing, and it's 10 times higher conversion than kind of just like roaming through throughout the app” |
| Yev's faceless YouTube Shorts channels | views One channel's views in the last 28 days, from its dashboard | 26M views/month | “So, this channel did 26 million views in the last 28 days. That's what this dashboard is showing, right?” |
| Yev's faceless YouTube Shorts channels | views Lifetime views on the showcased channel (ambiguous whether one video or the channel); another channel has even more | 1,000M views approximate | “That's the video, hundreds of millions of views, over like a billion views. And actually, another channel I have that's like even more” |
| Yev's faceless YouTube Shorts channels | count Faceless viral channels Yev runs (over 10) | 10 channels approximate | “But also, I have over 10 of these channels.” |
| benchmarkTikTok minimum ad budget per creative for Yev's outlier test | ad spend TikTok minimum per ad; can be switched off after ~$5 spend. Other platforms $5-$20 minimum | USD 20per creative approximate | “$20 minimum on TikTok, so then if we're talking about TikTok, I would go with example where you have to like put $20 in each one, but you could turn it off. After it spends $5, you could turn it off.” |