Muddy Waters’ recent short report on AppLovin reads serious. Abuse, violations, an impending takedown. But the discerning will realize something’s missing—like the stuff real short reports usually include. Somehow people are still treating it like they’ve found a dead unicorn in the server logs. See, the brilliance of the report isn’t the facts. It’s the lack of them. Just enough jargon and innuendo to let the market fill in the fear. And fear works fabulously here—we already hate ads. We already find them creepy. So when someone implies a high-performing ad platform is doing something unethical? Sounds about right. like the stuff real short reports usually include And fear works fabulously here—we already hate ads And fear works fabulously here—we already hate ads This isn’t a thesis. It’s a dare: “We bet you don’t understand this company well enough to call our bluff.” This isn’t a thesis. It’s a dare: “We bet you don’t understand this company well enough to call our bluff.” They’re not betting against AppLovin’s business, but against your comprehension. Classic intellectual arbitrage. They’re not betting against AppLovin’s business, but against your comprehension. They’re not betting against AppLovin’s business, but against your comprehension. Yes, Grok could disprove the paper. But as a marketer, I’m annoyed enough to go off on a rant of my own and explain how this stack works—and why none of their claims hold up. What Muddy Waters Is Actually Saying (In Plain English, With a Straight Face) What Muddy Waters Is Actually Saying (In Plain English, With a Straight Face) Here’s the report, minus the hedge fund perfume: AppLovin’s ads mostly go to people who were already going to buy. They use a tracking pixel that watches what people do on websites—and maybe that crosses a line with Apple or Google. Their AI engine (AXON) sounds impressive but probably isn’t doing anything special. Some ecomm advertisers tested the platform and didn’t come back. It smells like Cheetah Mobile—because why not throw in a fraudulent Chinese app that was banned, just for dramatic effect? AppLovin’s ads mostly go to people who were already going to buy. They use a tracking pixel that watches what people do on websites—and maybe that crosses a line with Apple or Google. Their AI engine (AXON) sounds impressive but probably isn’t doing anything special. Some ecomm advertisers tested the platform and didn’t come back. It smells like Cheetah Mobile—because why not throw in a fraudulent Chinese app that was banned, just for dramatic effect? That’s it. No cooked books, internal emails, or a “holy sh*t” revelation. Just a handful of what-ifs and “trust us, this is bad” vibes duct-taped together with technical vocab all compressed into an opportunistic PDF. Just a handful of what-ifs and “trust us, this is bad” vibes duct-taped together with technical vocab all compressed into an opportunistic PDF Just a handful of what-ifs and “trust us, this is bad” vibes duct-taped together with technical vocab all compressed into an opportunistic PDF It all sounds terrifying if you’ve never run an ad campaign. It all sounds terrifying if you’ve never run an ad campaign. But for all those words, not a single one was spent on the most basic question: What does this company actually do? But for all those words, not a single one was spent on the most basic question: What does this company actually do? What does this company actually do? Probably because once you do, most of their arguments fall apart. So let’s do it for them. So let’s do it for them What AppLovin Actually Does (Since Apparently That Wasn’t Relevant) What AppLovin Actually Does (Since Apparently That Wasn’t Relevant) AppLovin shows ads inside mobile apps—mostly games. You open a game, you watch an ad to get more coins—that’s them. Just infrastructure that makes in-app ads work. Just infrastructure that makes in-app ads work. Developers plug in the SDK (MAX) Developers plug in the SDK (MAX) Developers plug in the SDK (MAX) MAX powers the ads inside mobile games—rewarded videos, playables, interstitials—and automatically chooses the highest-paying one. It fills the space, runs the auction, and handles the logic. [Sidebar: How AppLovin Quietly Took Over 140,000 Apps] [Sidebar: How AppLovin Quietly Took Over 140,000 Apps] [Sidebar: How AppLovin Quietly Took Over 140,000 Apps] AppLovin didn’t try to out-Unity Unity. Unity built for prestige. Big studios. Serious games. AppLovin built for the ones Unity scoffed at—hypercasuals, solo devs, and the teams that just needed to make rent. Turns out there are a lot more of them. And MAX didn’t launch cold. AppLovin owned big studios, and pumped its own ad budgets through the system—spinning up liquidity before anyone else even showed up. Day one, it worked. That’s how you take over 140k apps without a parade: just pay people faster than anyone else. Advertisers bid through AppLovin Exchange (ALX) Advertisers bid through AppLovin Exchange (ALX) Advertisers bid through AppLovin Exchange (ALX) Mostly game studios and ecommerce brands. ALX pulls in demand from DSPs, programmatic buyers, and direct clients. Then AXON, the machine learning engine, decides which ad shows up, where, and for how much. The payout loop The payout loop The payout loop If a user taps, installs, or buys—the advertiser pays. AppLovin takes a cut. The rest goes to the developer. The better the ad performs, the more it gets shown. The more it gets shown, the more everyone earns. Most adtech stacks rent the data, piggyback the pipes, and take a cut on their way through. They sit between the buyer and the seller and hope no one notices they don’t do anything. AppLovin doesn’t rent. It runs. AppLovin doesn’t rent. It runs. It owns the SDK. The auction. The exchange. The optimization engine. It makes the decisions and delivers the results. That level of control is rare in adtech. But it’s not illegal. It’s what Apple, Amazon, and Google have done for years—because full control isn’t a red flag. It’s just what being better built looks like. But it’s not illegal. So no—AppLovin isn’t shady. But pretending full-stack execution is fraud? That might be. So no—AppLovin isn’t shady. But pretending full-stack execution is fraud? That might be. So no—AppLovin isn’t shady. But pretending full-stack execution is fraud? That might be. An Autopsy of a Report That Never Had a Pulse An Autopsy of a Report That Never Had a Pulse Great—now that we’ve done their homework for them, time to grade their report. Gloves off. “It’s Just Retargeting” — Cool, That’s Called Performance Marketing “It’s Just Retargeting” — Cool, That’s Called Performance Marketing One of the main claims in the Muddy Waters report is that over half of AppLovin’s conversions come from retargeting. They frame that like it’s a bad thing. Like AppLovin is cheating the system by showing ads to people who were already going to convert. Here’s the part they skip: retargeting is the part of marketing that gets people to pay. Here’s the part they skip: retargeting is the part of marketing that gets people to pay. Think about it: You’re running an ecomm brand, trying to acquire users profitably. If someone clicked your site, added to cart, then bailed—you want that person back. You don’t need 100% incrementality. You need ROI.And that’s exactly what AppLovin helps advertisers optimize for: time-to-purchase, payback window, LTV curves—not brand awareness. Direct-response, down to the decimal point. time-to-purchase, payback window, LTV curves—not brand awareness time-to-purchase, payback window, LTV curves—not brand awareness So when the report implies that retargeting = fake conversions, the only thing they’re exposing is that they know less about marketing funnels than your intern. the only thing they’re exposing is that they know less about marketing funnels than your intern. the only thing they’re exposing is that they know less about marketing funnels than your intern. “Their Pixel Might Violate Platform Rules” — Cute Story, Still No Evidence “Their Pixel Might Violate Platform Rules” — Cute Story, Still No Evidence Muddy Waters tries to paint AppLovin’s tracking pixel as a liability—implying it might violate Apple or Google policies and lead to a Cheetah Mobile-style takedown. They don’t show how. They don’t show that it has. They just raise the question and let paranoia do the rest. This part of the report is the classic hedge fund trick: Float the idea of a Terms of Service violation, backfill with vibes, then hope the market scares itself. Don’t let them terrify you: Every major ad platform uses pixels. That’s how you track conversions. AppLovin’s version isn’t weird—it’s standard. Unlike everyone else’s pixels, AppLovin’s never needed device IDs. It was built to work after Apple nuked IDFA. That’s the whole point of AXON: make smart predictions with limited data. There’s no sign of platform backlash. Apple and Google are not subtle when they don’t like you. If AppLovin were violating anything serious, they wouldn’t still be embedded in 270,000+ apps. They’d be gone. Every major ad platform uses pixels. That’s how you track conversions. AppLovin’s version isn’t weird—it’s standard. Unlike everyone else’s pixels, AppLovin’s never needed device IDs. It was built to work after Apple nuked IDFA. That’s the whole point of AXON: make smart predictions with limited data. There’s no sign of platform backlash. Apple and Google are not subtle when they don’t like you. If AppLovin were violating anything serious, they wouldn’t still be embedded in 270,000+ apps. They’d be gone. So unless Muddy Waters has an email from Tim Cook saying, “We’re coming for AppLovin,” this is just a vibe-based accusation with nothing underneath it. And worse, this argument completely misses the point. The pixel isn’t the advantage. It’s the input. The real edge—the reason AppLovin works—is AXON. And the fact that Muddy Waters barely mentions it tells you everything. And worse, this argument completely misses the point And worse, this argument completely misses the point And the fact that Muddy Waters barely mentions it tells you everything. AXON Isn’t Overhyped. It’s the Reason AppLovin’s Still in Business. AXON Isn’t Overhyped. It’s the Reason AppLovin’s Still in Business. Muddy Waters tries to make AppLovin’s pixel sound like a surveillance device. Relax. It’s just a signal. Relax The real story is AXON, which the firm conveniently brushes off as “overhyped AI”. Sure. And Google is just a search bar. Sure. And Google is just a search bar. Sure. And Google is just a search bar. AXON’s the entire reason AppLovin didn’t get body-bagged by Apple’s privacy changes. They didn’t have the biggest name or the most headlines. They were simply tighter, faster, and meaner. Here’s what gave AXON the edge: AppLovin controls the whole loop. The SDK handles the ad call. Their ad server runs the bidding. AXON saw the whole picture: who bid, who won, what ran, what worked. That’s the training data advantage everyone else wishes they had. They trained it in the most cutthroat ad environment there is: mobile games. Most of AppLovin’s ads run inside mobile games. That’s a ruthless environment: users skip garbage, and devs cut platforms that don’t monetize. And they didn’t just serve into that world—they owned a piece of it. That gave AXON full visibility: clicks, retention, spend, behavior—not just impressions, but outcomes. AXON wasn’t rebuilt. AXON was born after IDFA’s funeral. While everyone else was scrambling to patch their tracking-addicted systems, AppLovin skipped the panic and started clean. No duct tape. No rewrites. Just a head start—and a system that never needed surveillance to work. AppLovin controls the whole loop. The SDK handles the ad call. Their ad server runs the bidding. AXON saw the whole picture: who bid, who won, what ran, what worked. That’s the training data advantage everyone else wishes they had. AppLovin controls the whole loop. They trained it in the most cutthroat ad environment there is: mobile games. Most of AppLovin’s ads run inside mobile games. That’s a ruthless environment: users skip garbage, and devs cut platforms that don’t monetize. And they didn’t just serve into that world—they owned a piece of it. That gave AXON full visibility: clicks, retention, spend, behavior—not just impressions, but outcomes. They trained it in the most cutthroat ad environment there is: mobile games. AXON wasn’t rebuilt. AXON was born after IDFA’s funeral. While everyone else was scrambling to patch their tracking-addicted systems, AppLovin skipped the panic and started clean. No duct tape. No rewrites. Just a head start—and a system that never needed surveillance to work. AXON wasn’t rebuilt. It learned fast. It learned under pressure. And that pressure built a monster. So, AXON isn’t overhyped. It’s just uncomfortable for anyone who built slower or trained dumber. It’s just uncomfortable for anyone who built slower or trained dumber It’s just uncomfortable for anyone who built slower or trained dumber “Churn” Isn’t a Red Flag. It’s a Line in a Spreadsheet Without Context. “Churn” Isn’t a Red Flag. It’s a Line in a Spreadsheet Without Context. Muddy Waters cites a 23% churn rate among AppLovin’s ecommerce advertisers like it’s proof of product rejection. But here’s what they leave out: AppLovin only opened its platform to ecommerce in late 2024. These weren’t legacy clients backing out. They were net-new performance advertisers—most of them mobile-first for the first time—testing in-app ads with small, quarterly budgets. AppLovin only opened its platform to ecommerce in late 2024. AppLovin only opened its platform to ecommerce in late 2024. Other things Muddy decided to skip over: There’s no sign this churn mattered. Revenue kept growing. Margins held. Budgets didn’t vanish. If this was true attrition, it would show up in the top line. It didn’t. Performance advertisers behave this way by design. Especially in ecommerce. They test channels against CAC, reallocate fast, and pause without ceremony. That’s normal behavior—not rejection. The 23% stat itself is unlabeled. Is it spend-based? Account-based? Were these $1K test budgets or six-figure campaigns? The report doesn’t say—because context would ruin the punchline. There’s no sign this churn mattered. Revenue kept growing. Margins held. Budgets didn’t vanish. If this was true attrition, it would show up in the top line. It didn’t. Performance advertisers behave this way by design. Especially in ecommerce. They test channels against CAC, reallocate fast, and pause without ceremony. That’s normal behavior—not rejection. The 23% stat itself is unlabeled. Is it spend-based? Account-based? Were these $1K test budgets or six-figure campaigns? The report doesn’t say—because context would ruin the punchline. And here’s the most important part: 77% of those e-commerce advertisers stayed. This wasn’t a client walking out. It was a market walking in. And here’s the most important part: 77% of those e-commerce advertisers stayed. This wasn’t a client walking out. It was a market walking in. The Numbers Don’t Lie—Which Is Exactly Why the Report Doesn’t Focus on Them. The Numbers Don’t Lie—Which Is Exactly Why the Report Doesn’t Focus on Them. Every good short report starts in the same place: The numbers. You show where the revenue is overstated. Where the costs are buried. Where the cash flow is smoke and mirrors. That’s not what Muddy Waters did here. Instead, they went after pixels, churn, and speculation—while largely avoiding the one thing that’s actually measurable: How AppLovin performs as a business. Because here’s what the numbers say: $3.2 billion in ad revenue last year—up 75% $2.1 billion in free cash flow, not “adjusted EBITDA,” not projections—cash 33%+ net income margins Over $2 billion in share buybacks—executed while insiders held their shares $3.2 billion in ad revenue last year—up 75% $2.1 billion in free cash flow, not “adjusted EBITDA,” not projections—cash 33%+ net income margins Over $2 billion in share buybacks—executed while insiders held their shares That’s not a business in trouble. That’s a company with real operating leverage, compounding returns, and enough confidence to return capital at scale. If there were something rotten, you’d see it in the margins. If churn was a real problem, it would hit revenue. If the pixel theory had teeth, you’d see risk priced in. But none of that’s happening—because the business is healthy, efficient, and growing. So why avoid the numbers? Simple, they don’t support the story. They Talked About Pixels and Churn. Not the Part That Prints Money. They Talked About Pixels and Churn. Not the Part That Prints Money. Muddy Waters had a lot to say about pixels, churn, and scary-sounding AI. Know what they barely mention? Gaming. Know what they barely mention? Gaming. Gaming Which is convenient—because it’s the one thing that actually explains why AppLovin’s ad business is doing so well—and why it’s about to do even better. Which is convenient—because it’s the one thing that actually explains why AppLovin’s ad business is doing so well—and why it’s about to do even better. and why it’s about to do even better Here’s what every marketer already knows but Muddy didn’t bother to say out loud: Social CPMs are through the roof Tracking is garbage Everyone’s tired of getting stalked around the internet for a pair of shoes they already bought Social CPMs are through the roof Tracking is garbage Everyone’s tired of getting stalked around the internet for a pair of shoes they already bought So where does attention still convert? Games. So where does attention still convert? Games. Games Games are the last place where users actually choose to watch ads. Not because they’re “emotionally resonant.” Because they bribe you. Watch an ad, get a power-up. Watch two, get another life. And games aren’t just for kids. They’re for anyone with a phone and a short attention span. Because they bribe you 73% of Zynga players say in-game ads help them discover brands 69% of users in Asia prefer rewarded video over any other format And 58% say they actually feel more connected to those brands 73% of Zynga players say in-game ads help them discover brands 73% of Zynga players say in-game ads help them discover brands 69% of users in Asia prefer rewarded video over any other format 69% of users in Asia prefer rewarded video over any other format And 58% say they actually feel more connected to those brands And 58% say they actually feel more connected to those brands This isn’t a niche. It’s the next performance channel. And AppLovin didn’t catch the wave—they got there before it formed. As for Muddy, they’re not stupid enough to overlook this story. They just couldn’t tell it without proving themselves wrong. Final Thoughts: The Real Risk Isn’t Churn. It’s Comprehension. Final Thoughts: The Real Risk Isn’t Churn. It’s Comprehension. AppLovin’s biggest problem isn’t a couple advertisers closing their accounts. It’s being too good at something the market CBF to understand. It’s being too good at something the market CBF to understand. It’s being too good at something the market CBF to understand. And Muddy Waters didn’t miss that—they exploited it. They didn’t find a scam. They found complexity—and realized confusion was probably more profitable. And Muddy Waters didn’t miss that—they exploited it. They didn’t find a scam. They found complexity—and realized confusion was probably more profitable. realized confusion was probably more profitable. Now I don’t own a single dollar of AppLovin stock. But if this is what passes for a short thesis today, we’re not punishing fraud. We’re punishing competence.
