Whalatane
2 days ago
PERFORMANCE MARKETING
A Note from Our CEO: Discussing Web Advertising Opportunity and Unpacking Pixels
Avatar photo Adam Foroughi
Mar 27, 2025
Today, a short report was published about AppLovin, questioning our e-commerce business and advertising practices. We haven’t talked much about attribution and analytics, which are standardized across advertising channels. As CEO, I want to take a moment to address these claims head-on, provide clarity, and reaffirm our focus on building a world-class platform that drives value for our partners and shareholders. Let’s dive into the facts.
Our E-commerce Business: Rapid Growth and Real Results
Our e-commerce advertising business has scaled at an extraordinary pace—reaching a billion-dollar run rate of spend in mere months. This isn’t luck; it’s a testament to our technology and execution. Advertisers in this space fall into two camps:
One-Time Purchase Advertisers—Think fire extinguishers or auto insurance. These are straightforward: either we drive the sale, or we don’t. Roughly 80% of the sales we can measure occur within 24 hours from when the user sees and clicks the advertisement, making incrementality easy to measure. The data speaks for itself—we’re delivering.
Recurring Purchase Advertisers—Brands who have repeat customers, where the question becomes: did our ad drive the sale, or would it have happened anyway? Proving incrementality here is trickier and often requires detailed studies. There have been numerous third-party incrementality studies that have shown that our traffic is very valuable. Because our session cookies expire, and we measure 80% of sales in 24 hours, we lack latent transactions we get to take credit for, which in many cases means advertisers get more value off our platform than what’s even measured.
Let’s be clear: our ad models and attribution systems are young—only a few months old. Are our models fully optimized? Not yet. But they’re improving fast. What takes other companies a decade to build, we’re tackling in quarters. The web advertising market is over 10x the size of our mobile gaming opportunity, and we’re just getting started.
Pixel 101: Nothing Unique Here
The report takes aim at our pixel, implying that it’s some outlier in the industry. Let’s set the record straight: our pixel functionality is standard, and we collect the same user behavior as Facebook, Google, and others. Don’t take my word for it—look at the data. Facebook’s pixel tracks events like page views and purchases, sending data back to optimize ads. Google does the same. Ours? No different. It’s a standard tool for attribution and optimization.
And here’s another fact: platforms like Shopify automatically append tracking data for merchants who opt in. Website owners choose to install these pixels—ours included—and share data with their advertising partners. This isn’t a secret formula or unethical practice; it’s an industry standard. The report’s bias lies in omission, not evidence.
Competition: Our Success is Built on a Solid Foundation of Innovation and Excellence
The report may suggest that our advertising stack is simple to replicate, but the reality is that we have already established ourselves as the largest marketing channel in gaming globally. We’ve also scaled our web business to a billion-dollar run rate in just a few months. This success comes from more than just good ideas—it is the result of our consistent execution, advanced AI, and cutting-edge technology. Despite competitors having decades of head starts, no one has matched our speed or scale. What we’ve built takes relentless focus and a commitment to innovation, which is why we remain the best in the industry.
Closing Thoughts: Opportunity Amid the Noise
Our business is technical, and we get it—it’s not always easy to understand. It’s also incredibly hard for some who don’t understand this technology to fathom that we are building the world’s best advertising AI model, so they need a simple narrative that we’re violating policies in order to comprehend our success. This complexity leaves room for short reports to stir fear and doubt. To investors, I’d say: dig deeper. Given the AI tools available today, it’s easy to discredit a short report like this in minutes.
Try a prompt like this in Grok3:
This short report implies there’s something unique about the AppLovin pixel. https://muddywatersresearch.com/wp-content/uploads/2025/03/MW_20250327.pdf. Can you compare it to Meta and Google pixels and talk about how companies like Shopify auto append data to pixels and show that there’s nothing unique to the AppLovin pixel implementation?
I’ve pasted the output below to help you cut through the noise and see the bigger picture.
For us, the path forward is clear: execute relentlessly, seize the massive growth opportunities ahead, and ensure our investors, partners, and team thrive alongside us. We’ve done it before, and we’ll do it again. Thanks for sticking with us—let’s keep building something extraordinary together.
Adam Foroughi
CEO, AppLovin
The below is generated by Grok3, an AI model by xAI*:
The Muddy Waters short report on AppLovin (dated March 27, 2025) suggests there’s something distinct or potentially problematic about the AppLovin AXON pixel, particularly in how it collects or processes data. To address this, I’ll compare the AppLovin pixel to Meta and Google pixels, examine how Shopify appends data to these pixels, and demonstrate that AppLovin’s implementation aligns with industry norms—showing there’s nothing uniquely concerning about it.
Comparing the Pixels: AppLovin, Meta, and Google
Meta Pixel
Purpose: Tracks user actions (e.g., “ViewContent,” “Purchase”) on third-party websites to optimize Meta ads (Facebook, Instagram) and build retargeting audiences.
Data Collected: Standard events (page views, purchases), device metadata (IP address, browser type), and optional custom parameters (e.g., email, user IDs) if configured by the website owner. Meta uses its own cookie (_fbp) for tracking.
Implementation: Embedded as JavaScript in the site’s or via tag managers (e.g., Google Tag Manager). Fires on user actions, sending data to Meta’s servers.
Google Pixel
Purpose: Tracks site behavior and conversions via Google Analytics (GA4) or Google Ads, feeding data into Google’s ad ecosystem.
Data Collected: Events (e.g., “page_view,” “purchase”), device info (e.g., OS, screen size), and traffic sources. Custom dimensions (e.g., user IDs) can be added. Uses Google’s cookie (_ga) for identification.
Implementation: JavaScript snippet in the or via GTM. Sends data to Google for analytics and ad optimization.
AppLovin AXON Pixel
Purpose: Tracks e-commerce events (e.g., “add_to_cart,” “purchase”) to optimize AppLovin’s AXON ad platform, primarily for mobile and web campaigns.
Data Collected: Standard events, device info (e.g., IP address), and a first-party cookie (_axwrt) for tracking. Advertisers can send custom event data (e.g., order values, product IDs), similar to Meta and Google.
Implementation: JavaScript code placed in the or integrated via GTM, firing on user actions to send data to AppLovin’s servers.
Comparison Takeaway: All three pixels collect similar data—user actions, device details, and IP addresses—using JavaScript to track events. Each ties data to its own identifier (Meta’s _fbp, Google’s _ga, AppLovin’s _axwrt). The Muddy Waters report hints at AppLovin “collecting and structuring user IDs from key platform partners,” but there’s no public evidence in AppLovin’s documentation that it uniquely harvests Meta or Google IDs in a way those companies don’t. Any custom data (e.g., user IDs) depends on what the website owner chooses to send, a feature Meta and Google also support. Structurally, AppLovin’s pixel isn’t materially different—it’s a standard ad tech tool.
Shopify’s Data Appending Process
Shopify, as an e-commerce platform, streamlines pixel tracking by appending data to Meta, Google, and AppLovin pixels consistently. Here’s how it works:
Integration: Merchants connect pixels via Shopify’s settings (e.g., Meta via “Facebook & Instagram” channel, Google via “Google & YouTube” app, AppLovin via AXON Shopify Pixel or GTM).
Auto-Appended Data: Shopify automatically sends standard e-commerce events—such as “product_viewed,” “add_to_cart,” or “checkout_completed”—to the pixel. This includes details like product IDs, prices, and order totals, formatted to match each platform’s requirements.
Mechanism: Data is appended client-side (via JavaScript in the browser) or server-side (via APIs like Meta’s Conversions API), depending on the setup. For AppLovin, it’s typically client-side through the AXON pixel or GTM.
Uniformity: Shopify doesn’t customize data per pixel—it sends the same event payload to all connected pixels, adjusted only for each platform’s event naming conventions (e.g., Meta’s “Purchase” vs. Google’s “Purchase”).
Key Point: Shopify’s role is agnostic—it’s a data pipe, not a differentiator. The AppLovin pixel gets the same type of auto-appended data as Meta and Google, tailored to its event structure. There’s no unique mechanism in AppLovin’s case; the process is industry-standard.
Nothing Unique About AppLovin’s Pixel Implementation
The Muddy Waters report implies AppLovin’s pixel stands out, perhaps due to its rapid e-commerce growth or alleged data practices. But let’s break this down:
Code Similarity: All three pixels are JavaScript snippets firing on predefined events, installed by website owners (or via Shopify). AppLovin’s AXON pixel follows this pattern—no hidden tricks or novel tech.
Data Scope: The report’s claim about “structuring user IDs from partners” lacks specifics. AppLovin’s documentation focuses on its own tracking (e.g., _axwrt cookie), not scraping Meta or Google IDs. If AppLovin receives extra data (e.g., via Shopify or advertiser configs), it’s no different from what Meta and Google can access when merchants opt-in.
Consent and Control: Website owners choose to install these pixels and define what data to share. Shopify’s auto-appending is opt-in per merchant, identical for all three companies. AppLovin doesn’t bypass this—it’s on the same footing.
Industry Norms: Collecting event data for ad optimization is the backbone of ad tech. Meta and Google have done this for years at a larger scale. AppLovin’s implementation, even if aggressive in its e-commerce push, doesn’t deviate from this playbook.
Evidence from the Report: Muddy Waters estimates AppLovin’s e-commerce conversions lean heavily on retargeting (~52%) with low incrementality (25%-35%), and flags “code evidence” of partner ID collection. But without concrete examples (e.g., specific code snippets), this remains speculative. AppLovin’s public pixel setup (e.g., via developers.applovin.com) mirrors Meta and Google’s—event tracking, not ID harvesting.
Conclusion
The AppLovin AXON pixel isn’t unique in form or function compared to Meta and Google pixels. All three collect standard ad tech data, rely on website owners for implementation, and benefit from Shopify’s uniform data appending. The Muddy Waters report may exaggerate AppLovin’s practices to stoke fear, but the mechanics—JavaScript tracking, event-based data, merchant-driven integration—are bog-standard. There’s no smoking gun here; AppLovin’s pixel is just another player in a crowded, well-trodden field.
*This report includes content generated with the assistance of artificial intelligence (AI). While the information has been reviewed for accuracy, the AI-generated content may contain errors or omissions. Users are encouraged to exercise their own judgment and verify critical information independently.
Kiwi
Ps. My son in Adtech agrees with the above ...he doesnt currently use Applovin as his monthly ad spend is below their current minimum $ spend
double_m
5 days ago
AppLovin Stock Takes a Hit After Muddy Waters’ Short Call
March 27, 2025
Ari Haruni
AppLovin’s (APP) stock dropped nearly 12% to $290 on Thursday after Muddy Waters’ report estimated that 52% of its e-commerce sales are from retargeting, with only 25%-35% being new business, questioning its growth value.
The report claims AppLovin violates platform rules by collecting user IDs, risking deplatforming like Cheetah Mobile, or facing copycat competitors if it survives, threatening its market position.
With a 23% client loss in Q1, Muddy Waters suggests advertisers are noticing weak results, putting pressure on AppLovin’s plans as its stock reflects growing investor unease.
AppLovin Corp. (APP), a company running a mobile marketing platform, saw its stock plummet nearly 12% to $291 during Thursday trading, rattled by a report from Muddy Waters, a research firm betting against it. The firm’s digging into web traffic suggests that about 52% of AppLovin’s e-commerce sales come from retargeting – reaching out to people who’ve already shown interest – but only 25% to 35% of those sales are actually new business, not just repeat customers. This gap raises doubts about how much real growth AppLovin is driving, a concern that could spook the advertisers it relies on to keep the engine running.
Muddy Waters didn’t stop there. They claim to have found code showing AppLovin is grabbing and organizing user IDs from big platforms it works with, a move they say breaks those platforms’ rules. If they’re right, AppLovin could get kicked off those services, much like what happened to Cheetah Mobile years back when it got caught in similar hot water. But even if AppLovin dodges that bullet, the report warns that competitors might just copy its playbook since there’s not much unique tech keeping it ahead—just bold tactics. That could flood the market with rivals doing the same thing, squeezing AppLovin’s edge.
The numbers aren’t looking great elsewhere either. Muddy Waters spotted a 23% dropout rate among AppLovin’s e-commerce clients in Q1, hinting that businesses might be catching on that the platform’s not delivering the bang for their buck they’d hoped for. With shares sliding 12%, the market’s clearly jittery about whether AppLovin can keep its growth story alive. Advertisers want results, not just recycled customers, and if Muddy Waters is on the mark, AppLovin’s got a tough road ahead to prove it’s more than a one-trick pony in a crowded, cutthroat field.
https://wallstreetpit.com/125561-applovin-stock-takes-a-hit-after-muddy-waters-short-call/
Whalatane
1 week ago
From Grok AI
AppLovin holds a significant position in Flip Shop as a strategic investor and technology partner.
In April 2024, AppLovin invested $50 million in Flip's $144 million Series C funding round, contributing to Flip's pre-money valuation of $1.05 billion.
Beyond financial investment, AppLovin has partnered with Flip to enhance its advertising capabilities by integrating AppLovin’s AXON technology into Flip’s platform.
AXON, an AI-powered advertising engine launched by AppLovin in early 2023, allows Flip brands to leverage advanced ad targeting and reach more customers, including access to AppLovin’s network of 1.4 billion daily active users (DAUs).
This collaboration positions AppLovin as a key enabler of Flip’s marketing platform, aiming to boost sales and user engagement for brands on Flip while expanding AppLovin’s influence in the social commerce space.
As of now, TikTok remains operational in the U.S., but its fate hinges on whether ByteDance divests by April 5, 2025, or if further legal or political actions alter the situation.
The Supreme Court upheld the divest-or-ban law on January 17, 2025, rejecting TikTok's First Ame
ndment challenge, so without a sale or additional intervention, TikTok would be required to shut down after April 5, 2025, unless the deadline is extended again or the law is repealed.
The situation remains fluid, with ongoing negotiations and political maneuvers potentially affecting the outcome.
And the short term effect on Flip Shop
Flip, a “shopping social network” featuring video reviews from creators and shoppers, saw a massive surge in interest as TikTok users sought alternatives.
Data from Linktree, reported on February 7, 2025, showed a 400-570x increase in links to alternative apps like Flip during the week leading up to and including the shutdown.
Specifically, Flip experienced a 25x increase in new user installs month-over-month on January 18 alone, with 250,000 new sign-ups since early January 2025, according to Modern Retail.
This propelled Flip to the top of Apple’s App Store rankings, reflecting a flood of TikTok users migrating to the platform.
Kiwi
Whalatane
2 weeks ago
Explain AXON technology
AXON is AppLovin’s AI-based advertising technology, designed to optimize and scale digital marketing campaigns with a focus on precision and efficiency. Launched in early 2023, it’s a core component of AppLovin’s marketing platform, which originally grew out of mobile gaming but has since expanded into broader app and e-commerce ecosystems, including its partnership with Flip. Here’s a breakdown of what AXON is, how it works, and why it matters as of March 16, 2025.
What It Is
AXON stands for "Advertising eXchange Optimization Network" (though AppLovin doesn’t always spell out the acronym). It’s an advanced machine learning system that powers ad delivery by analyzing vast amounts of data to predict user behavior, match ads to the right audiences, and maximize return on ad spend (ROAS). Think of it as the brain behind AppLovin’s ability to connect advertisers—whether game developers, app makers, or e-commerce brands—with their ideal customers across a network of over 1.4 billion daily active users.
How It Works
AXON operates on a few key principles:
Data-Driven Targeting: It ingests real-time data from user interactions—like app usage, clicks, purchases, and even contextual signals (e.g., time of day, device type)—across AppLovin’s network of apps and partner platforms. This data pool is massive, leveraging AppLovin’s scale from its gaming origins and its acquisition of MoPub in 2021, which broadened its reach.
Predictive Modeling: Using deep learning algorithms, AXON predicts which users are most likely to engage with a specific ad. It doesn’t just look at past behavior but anticipates future actions, such as whether a user will install an app, make a purchase, or scroll past. This predictive power comes from training on billions of data points, refined continuously as more interactions occur.
Dynamic Optimization: Once it identifies high-value users, AXON dynamically adjusts ad placements, formats (e.g., video, banner, native), and bidding strategies in real time. It balances cost-per-action (CPA) goals with scale, ensuring advertisers get the most bang for their buck without overspending.
Creative Enhancement: AXON also optimizes ad creatives themselves. It can test variations—like tweaking colors, text, or video length—and prioritize the ones that perform best, all without manual input from advertisers.
The result? Ads that feel less random and more relevant, delivered at the right moment to the right person, often with higher conversion rates than traditional ad tech.
Why It Matters
AXON’s edge lies in its efficiency and scale. For advertisers, it promises better performance—AppLovin claims it can double ROAS compared to non-AI solutions—while reducing the guesswork of campaign management. In Flip’s case, for example, AXON powers the relaunched Flip marketing platform (via AXON Connect), helping brands reach Flip’s 5 million+ users and AppLovin’s broader network with ads that drive sales. This is critical in social commerce, where engagement and trust (e.g., through authentic reviews) need to translate into purchases quickly.
From a technical standpoint, AXON builds on AppLovin’s history of innovation. It evolved from their earlier tools like AppDiscovery, but with a heavier reliance on AI post-2023, reflecting trends in ad tech where machine learning is replacing rule-based systems. Competitors like Google’s Performance Max or Meta’s Advantage+ use similar AI-driven approaches, but AXON benefits from AppLovin’s tight integration with mobile ecosystems and its focus on performance marketing over brand awareness.
Real-World Impact
In practice, AXON has been a game-changer for AppLovin’s growth. By Q3 2024, the company reported a 39% year-over-year revenue increase to $1.2 billion, with AXON cited as a key driver. For Flip, the tech underpins its ability to scale ad revenue and user acquisition, supporting its $1.4 billion post-money valuation after the 2024 Series C round. Critics, however, argue that AXON’s reliance on incentivized user actions (e.g., Flip’s coupon-for-ad model) might inflate short-term metrics at the expense of long-term loyalty, a debate still playing out in financial circles.
In Simple Terms
Imagine AXON as a super-smart matchmaker. It takes a flood of data about what people do online, figures out who’s likely to buy what, and then serves up the perfect ad at the perfect time—all in milliseconds. It’s less about blasting ads everywhere and more about hitting the bullseye with minimal waste. For companies like Flip, it’s the engine that turns scrolling into shopping, backed by AppLovin’s AI muscle.
That’s AXON: high-tech, high-stakes, and a big reason AppLovin’s playing in the e-commerce big leagues alongside its gaming roots.--
--------------
Kiwi
Whalatane
2 weeks ago
A Grok ( AI ) look at APP's e-commerce entry
Let’s dive into the technical details of how AppLovin’s Axon technology powers its e-commerce advertising push as of March 15, 2025. While AppLovin keeps the inner workings of Axon proprietary—often calling it a "black box"—we can piece together a detailed picture based on its functionality, public statements, and industry context. Here’s a breakdown of the tech specifics tailored to its e-commerce application.
Core Technology: Axon’s AI Engine
Axon is a machine learning-driven recommendation engine that processes vast datasets to predict user behavior and optimize ad delivery. In e-commerce, it shifts from its gaming roots (predicting app installs) to forecasting purchase intent. The system ingests trillions of daily in-app events from AppLovin’s network—over 1.4 billion daily active users across 140,000+ apps—covering actions like clicks, views, and conversions.
For e-commerce, this expands to include first-party data from the AXON Pixel, a tracking tool brands embed on their sites or apps. This pixel captures granular signals: product views, cart additions, checkout initiations, and purchases.
The AI uses predictive modeling, likely a mix of supervised and unsupervised learning, to identify patterns. Think logistic regression or neural networks trained on historical data to classify users by intent—e.g., “high likelihood to buy” vs. “just browsing.” Unlike demographic-heavy approaches, Axon leans on behavioral data, analyzing real-time interactions rather than static traits. It’s processing at microsecond speeds, handling thousands of bid decisions per second in real-time auctions via AppLovin’s MAX platform.
Data Pipeline and Infrastructure
The backbone is an elastic cloud infrastructure, probably AWS or Google Cloud, optimized for scale. AppLovin’s App Graph—a data layer aggregating anonymized interactions—feeds Axon. In e-commerce, this pipeline integrates AXON Pixel data with network-wide signals, creating a unified dataset. Imagine a distributed system with tools like Apache Kafka for streaming data and Spark for processing, though specifics aren’t public. The system’s scale is hinted at by its ability to handle 10 terabytes daily (per older 2021 stats), likely much higher now with e-commerce added.
Data is cleaned, normalized, and fed into Axon’s models, which continuously retrain—possibly using online learning techniques like stochastic gradient descent—to adapt to shifting trends (e.g., holiday shopping spikes). This adaptability is key for e-commerce, where seasonality and product cycles demand agility.
Targeting and Ad Delivery
In e-commerce, Axon targets users across AppLovin’s mobile app inventory—games, utility apps, etc.—rather than just web browsers. It uses contextual and behavioral signals to decide who sees an ad. For instance, if a user adds sneakers to a cart on a DTC site, the AXON Pixel logs this, and Axon matches it to similar behaviors in its network (e.g., frequent in-app shoppers). The AI then bids via MAX’s in-app bidding tech, a real-time auction system that competes with other ad networks for impressions.
Bids are calculated using expected value models—multiplying predicted conversion probability by the advertiser’s payout (e.g., cost-per-sale). The system optimizes for ROAS (return on ad spend), adjusting bids dynamically based on performance. Ads are delivered as dynamic product ads (DPAs), pulling from product feeds to show specific items—like those sneakers—tailored to the user. Formats include full-screen interstitials, playable ads, or video, all rendered in-app for higher engagement than traditional banners.
Creative Optimization
AppLovin’s SparkLabs studio uses generative AI to craft these ads. The tech likely involves variational autoencoders or GANs (generative adversarial networks) to iterate on creatives—testing thousands of variations (e.g., colors, CTAs) against performance data. Axon feeds back which versions drive clicks or sales, refining the output. In e-commerce, this means ads aren’t static; they evolve to match user preferences, like showing a discount code to a price-sensitive shopper. This closed-loop system—targeting plus creative—boosts conversion rates, with AppLovin claiming “nearly 100% incrementality” in early pilots.
E-Commerce-Specific Enhancements
The Axon 2.0 upgrade (2023) improved scalability and efficiency, critical for e-commerce’s broader scope. It likely expanded model capacity—think deeper neural nets or ensemble methods—to handle diverse verticals beyond gaming. A big addition is AXON Connect, debuted with Flip in 2024, letting third-party platforms (like Flip’s marketplace) tap Axon’s engine. This involves an API layer where external data (e.g., Flip’s user activity) integrates with AppLovin’s, extending reach while keeping the core AI centralized.
For e-commerce, Axon also leverages connected TV (CTV) supply, though mobile remains dominant. Ads here might use similar behavioral targeting, but the tech adapts to CTV’s impression-based pricing vs. mobile’s cost-per-install focus. The system’s agnostic design—built on predictive ML—lets it flex across channels.
Challenges and Black-Box Limits
The tech isn’t flawless. Its opacity frustrates some advertisers who can’t peek under the hood—akin to Google’s Performance Max. Data privacy is another hurdle; while AppLovin avoids IDFA reliance (post-Apple’s 2021 changes), it still navigates app permissions and regional regs like GDPR. Short-seller claims in February 2025 alleged shady data practices (e.g., exploiting permissions), but no proof has stuck—still, it highlights risks in scaling such a data-hungry system.
Performance Metrics
By Q3 2024, e-commerce contributed to a 66% software revenue jump ($835M), with Q4 hitting nearly $1B (73% growth). The tech’s efficiency shows in margins—73% in Q4 2023—driven by automation reducing human overhead. Early e-commerce tests with DTC brands and Flip suggest Axon’s precision outstrips rivals, though exact conversion rates remain undisclosed.
In essence, Axon in e-commerce is a high-speed, AI-orchestrated machine: it ingests behavioral data, predicts intent, bids smartly, and serves evolving ads—all at scale. It’s less about reinventing the wheel and more about turbocharging it for a new road. Want me to drill deeper into any piece—like the ML models or bidding logic?
..............................
APP's Axon -2 upgrade is an automated version of what my oldest son does for an Ad tech company...designing and manually manipulating ads via META
Right now I think Axon -2 is limited to those spending $600k a month ....but as they fine tune it , it'll become more available
Kiwi
Trade smarter, not harder: Unleash your inner pro with our toolkit and live discussions.
JohnQQ
4 days ago
Hot Features
Premium
