Muddy Waters short it too! Is Applovin really a "egg with seams"?

It's chaotic; in just three months into the new year, there have been five public short reports from different institutions. Who else can claim such "treatment"?

At the end of trading on Thursday Eastern Time, $ AppLovin.US faced a surprise short report from Muddy Waters. After being repeatedly shorted, investor confidence weakened, and with market sentiment fluctuating due to tariffs, Applovin bulls once again surrendered, with the stock price dropping nearly 20%.

There are some disputes between the bulls and bears regarding the points raised in the Muddy Waters report, which we will discuss in detail later. However, as some bulls are concerned, the increasing number of short reports, while not proving that Applovin is completely fraudulent (non-financial fraud cannot be pinpointed immediately, mostly involving risks on the edges of business and compliance), does not lead to the hoped-for judgment and defeat.

However, the rumored mysterious technological black box, being gradually exposed and unboxed, may lead the market to "disenchant" Applovin's technological content, thereby suppressing the valuation ceiling granted by AI concepts, including the space for stock price rebounds.

Especially concerning the potential violations mentioned by Muddy Waters regarding Applovin's collection of user data to build a PIG database, this may attract the attention of regulatory authorities. This is also a more detailed description of the "theft of Metadata" issue raised by Fuzzy Panda last time.

However, this time, Applovin's management reacted much faster to the shorting. The CEO quickly responded, denying the allegations, and hired a professional investigation agency to investigate the recent series of sporting events. An independent investigation report will be issued later, and Dolphin Research will follow up on this.

So, what new logic does this Muddy Waters short report bring? What kind of impact does it have on Applovin?

1. "Smart" Muddy Waters

Muddy Waters is well-known not only on Wall Street but also in the Chinese investment circle. This is because many of the companies they short are Chinese concept assets, such as Huishan Dairy, Focus Media, Luckin Coffee, and Anta.

Although Muddy Waters has a history of short cases with a not low probability of being wrong, this short report also has some obvious bugs (typos, mismatched data timestamps, insufficient research, etc.), which are used by Applovin bulls to question their professionalism. However, in Dolphin Research's view, Muddy Waters is much smarter than the previous ones.

Lauren, Bear Cave, Cluper, and Fuzzy Panda have a high proportion of discussions in their short views regarding Applovin's game advertisements inducing users to download, leading to inflated advertising effectiveness (click-through rates), which serves as their main weapon of attack But Dolphin Research also mentioned that while there is ample evidence for this, it is not the biggest concern, especially for the gaming clients in the case presented.

This is because, in a commercial environment, the client (Party A) generally measures the value of an investment based on results, specifically looking at the final ROI, rather than relying solely on some easily manipulated metrics during the process, especially those "operational performance indicators" provided by Party B. Otherwise, this numerical game would not last more than a year at most.

Therefore, most previous gaming clients were aware of Applovin's "small actions"—as long as your small actions do not significantly affect my final accounting, I can turn a blind eye to these discrepancies.

However, the cleverness of Muddy Waters lies in the fact that it did not repeatedly emphasize the fraudulent performance indicators of historical game advertisements, but instead directly targeted Applovin's current growth valuation support point—e-commerce advertising. In simple terms, the entire report attempts to prove:

Applovin used illegal and unfair competition methods to obtain and collect e-commerce user data from other platforms, achieving high conversion rates through repeated targeting rather than AI. Therefore, the actual incremental effect of converting users is far from what the CEO claimed, and after just one quarter, the interest of e-commerce clients has already weakened.

If the charges of illegally and unfairly obtaining user data are substantiated, the worst outcome would not only be that peers like Meta and Google, whose data was stolen, forcibly cut off potential ties with Applovin but there could also be a ban imposed by Apple on Applovin.

2. Evidence has loopholes, but the impact is significant

So how did Muddy Waters present the facts and list the evidence? Dolphin Research categorized it into two points, with the first point being the most impactful:

(1) The biggest weapon: Commercial risk becomes regulatory risk

Muddy Waters pointed out that Applovin illegally obtained and tracked e-commerce user data, violating Apple's user privacy agreement. Additionally, since it does not have a ToC platform, the acquisition of this e-commerce user data also comes from intercepting data from peers. Muddy Waters detailed in the report how Applovin "stole data" and repeatedly targeted "high-quality users" with existing shopping intentions, thereby achieving higher conversion efficiency than its peers.

This issue was previously raised by Fuzzy Panda (searching for a product keyword on Facebook and then seeing related product ads from Applovin in-game ads), but the core point is how Applovin obtained user IDs from third-party platforms. Muddy Waters' report provided some explanation for this.

There are two key steps in the entire process:

a. Filtering out high-quality users' third-party IDs

Fuzzy Panda previously mentioned the data theft issue involving Applovin and Meta—Applovin had clients run a certain scale of ads on Meta, and Meta returned advertising performance data to the advertisers, which included user IDs (such as fbp on Facebook, igld on Instagram) and detailed device information Perhaps it is this method that requires customers to spend an extra amount of money, so Applovin does not solely rely on this channel. Muddy Waters listed another data acquisition method, which is that Applovin inserts its Applovin Pixel (an attribution tool for the advertising platform, generally used to track advertising effectiveness and improve conversions) into the advertisers' official websites to track user behavior on the official site.

When a user with purchase intent accesses the product's official website through a third-party platform, such as Facebook ads, adds items to the shopping cart, and initiates checkout, it triggers the Applovin Pixel to output the user's information, including Facebook ID and device information that can be authorized by the product's official website.

Similarly, if the merchant uses the e-commerce hosting and payment settlement services provided by Shopify, the Applovin Pixel will also return Shopify's cookie data (Wrt3p shopping cart event ID and Y cookie user session ID), making the user filtering even more precise.

b. Connecting third-party IDs through device fingerprinting information

Subsequently, for these filtered user IDs (from different platforms), Applovin compares device information with what it has collected (from a large number of gaming applications or directly from mobile OEM manufacturers) and then fits its designated user ID with the third-party platform's user ID (such as the aforementioned Facebook ID and Shopify ID) to locate a specific user.

We know that in Europe and the United States, users' personal identity information is strictly protected; otherwise, it can easily be restricted by large platforms or targeted by regulatory authorities. Therefore, both device manufacturers and app platforms assign a unique identifier ID to users. Although this ID does not contain personal identity information, it allows for continuous tracking of behavioral data under this unique ID.

However, different platforms assign different IDs to the same user. If they can be merged and aligned, even though it is unknown which person (identity information) this data belongs to in the real world, the platform can locate "a specific person" based solely on this ID, enabling precise advertising targeting, etc.

The process of merging and aligning different IDs cannot do without "Fingerprinting" device fingerprint information Device fingerprint information refers to device parameter values that are not easily changed by users over a long period, such as hardware information (device model, operating system, device resolution, etc.), network parameters (IP address, time zone, etc.), and advanced device settings (memory usage, boot time, font, etc.), which are used to form a unique device identifier for specific users. This identifier enables cross-application user behavior tracking.

To some extent, it can be said that this is equivalent to linking the user IDs of different platforms. In this way, Applovin can use its cross-platform unified identifier to continuously track user behavior, ultimately forming a PIG database (Persistent Identity Graphs, PIGs) that can directly locate users over the long term, thereby enhancing its advertising targeting and conversion capabilities.

The above process can be done by Meta or Shopify, whether it is Meta tracking Shopify or Shopify tracking Meta. However, due to their direct focus on end-users and large traffic, these platforms face stricter legal supervision and peer monitoring when obtaining data from third-party platforms.

For example, Apple's user privacy agreement explicitly requires applications not to collect user device information in violation of regulations, or they risk being removed from the platform.

However, the cunning part here is that at the beginning of the year, Applovin sold off its only 2C applications—casual games and the game studios behind them. From the perspective of direct responsibility, Apple's removal warning seems to have less constraint on Applovin.

Nevertheless, this contradicts the CEO's statement that "we have not viewed user data from other companies..."

In addition, there is indeed a ticking time bomb here; as Applovin's scale gradually expands in the future, the probability of being monitored by regulators will also increase, thereby limiting Applovin's growth space.

Dolphin Research also mentioned in a previous bearish report that when the scale is small, competitors and regulators may temporarily set aside issues, but once the scale reaches a certain level (for example, e-commerce revenue reaches 5 billion USD), it can easily lead to scrutiny. At that point, it will depend on whether Applovin can clean up its act before the data accumulation is sufficient to bring about a technological upgrade.

(2) E-commerce advertising effects are exaggerated, and customers are churning

This point has little incremental information, and Dolphin Research also has some reservations In simple terms, Muddy Waters believes that Applovin's effectiveness in converting users through e-commerce advertising is not as the CEO stated in the 3Q24 conference call, claiming a 100% increase, but rather between 25% to 35%. This is based on their research of data from five Shopify platform merchants (covering approximately 37 million users). Furthermore, due to some inherent repurchase intent from users, the actual increase may be even lower.

At the same time, Muddy Waters tracked that among the 776 e-commerce websites participating in Applovin's e-commerce advertising test, many removed the Applovin Pixel after three months, indicating dissatisfaction with the effectiveness of Applovin's ads to some extent.

The evidence presented by Muddy Waters indicates that Applovin's path in the e-commerce sector is not smooth sailing. However, it is common for clients to exit due to dissatisfaction in the early stages of testing, and this does not rule out the possibility that Applovin may regain client interest through continued adjustments and optimizations.

Moreover, the five merchants selected by Muddy Waters for research have timestamp discrepancies in the screenshot evidence provided, and while quantitative conclusions were drawn, no quantitative evidence was offered, making it difficult to fully reference. It is also unclear whether there is a significant overlap between the target users and Applovin's gaming users (the brand names or product categories of the five merchants were not specified).

However, it must also be acknowledged that there are indeed concerns that Dolphin Research has consistently mentioned—due to the lack of first-party e-commerce user data advantages, Applovin will need to undergo a "cultivation period" similar to what it experienced in the gaming sector, and AI may help accelerate this process. But a skilled cook cannot make a meal without rice; since Applovin cannot obtain first-party data from a ToC platform as it did with gaming studios, and given that e-commerce advertising places more emphasis on actual GMV transaction effectiveness, Applovin does not possess a competitive advantage. The reason we previously emphasized not being overly pessimistic in the early stages is simply because:

a. The e-commerce advertising market is large enough that even a small penetration by Applovin can significantly boost current performance. For example, the estimated revenue for Q4 is between $50 million to $100 million, and the CEO disclosed a trend and capability for an annualized revenue scale of $1 billion this year.

b. Applovin's vast gaming user data, which skews towards a more female and older demographic, overlaps with shopping user characteristics. In a context where advertising prices on traditional channel platforms are relatively high (after the iOS privacy policy changes, more and more small to medium advertising platforms have lost their competitive edge, leading advertisers to shift budgets to major platforms like Meta, Google, and Amazon over the past three years, which has temporarily inflated eCPM due to overcrowding), Applovin's cost-effectiveness advantage can still attract some clients.

The above two points of logic have not yet been completely disproven by the short report, but they may weaken bullish sentiment and confidence in the short term Overall, Muddy Waters' report is more focused on the key issues than previous reports, as it targets the e-commerce advertising growth prospects of Applovin, which are inherently weak and have not yet fully gained customer validation and trust. Although Applovin's small tricks are common in Greater China, regulatory bodies in Europe and the United States are very active, and users are more concerned about privacy violations, constantly being monitored cannot withstand the pressure.

If we must take an extremely pessimistic scenario, eliminating the growth of e-commerce advertising, then according to the management's guidance in the fourth-quarter conference call, the traditional gaming business can maintain a 20% organic growth rate. Therefore, Dolphin Research conservatively values it at $70 billion (only considering the advertising part, with a 2025 EV/EBITDA of 22x), slightly higher than the market value before the third-quarter report release ($58 billion).

Dolphin Research "Applovin" Historical Study

Financial Report

February 13, 2025 Conference Call Minutes: Applovin (Summary): The Best Case of Technology Meeting Cross-Market Demand

February 13, 2025 Financial Report Commentary: Surging 30%! How Did Applovin Crush the Shorts?

In-Depth

January 10, 2025 First Coverage Part 2: Following the Pattern, Can Unity Replicate Applovin's "Cash Ability"?

January 3, 2025 First Coverage Part 1: The "Revealing" Applovin: A Winning Game Planned for Five Years

Hot Topics

February 28, 2025 Short Report Commentary: Heavily Discounted, Is Applovin Guilty of This?

Risk Disclosure and statement of this article: Dolphin Investment Research Disclaimer and General Disclosure