Sensor Tower and Mobile Index Revenue & User Estimation Methods
As competition in the mobile app market intensifies, developers and investors rely heavily on data analytics platforms to understand market trends and make strategic decisions. User and revenue estimates provided by platforms like Sensor Tower and Data.ai are widely used as industry-standard metrics. This document aims to analyze the methodologies by which these platforms estimate key metrics such as user count and revenue, highlighting the basis for their reliability and their inherent limitations.
Data Basis for User Estimation
Platform user estimates are based on data that can identify devices or users.
- ADID (Android Advertising ID): A 36-digit unique identification code used on Android devices
- IDFA (Identifier for Advertisers): A unique identification code used on iOS devices
In addition to these identifiers, platforms acquire data through paths unknown to users to add depth to their analysis. Representative examples include free VPN services, reward apps that offer compensation for using specific apps, or utility apps directly distributed by data collection companies. These apps collect user data in exchange for providing free services, which form a component of the analytics platforms' data pool.
Revenue Estimation Methodology and Modeling
Revenue estimation is conducted by statistically modeling revenue ranking data publicly available in app markets. Some app developers provide paid viewer access for their projects to data analytics platforms (data barter), allowing the platforms to access actual revenue data recorded in the Google Play Console or App Store Connect. Based on the data secured in this way, the platforms build correlation models showing that a specific revenue rank corresponds to a specific revenue scale. For example, by learning from the data of multiple apps, they establish criteria such as "$1 million for 1st place" and "$100,000 for 10th place." Afterwards, they expand and apply this model to the entire app ecosystem, inversely calculating each app's revenue based solely on its ranking.
Limitations of Estimation Models and Considerations for Interpretation
While this estimation method allows for convenient revenue estimation, it has structural limitations that require a critical perspective when interpreting the data. Because most analyses focus on Google Play and Apple App Store, revenues from specific app markets like the Galaxy Store or other platforms like Steam and PlayStation are excluded from the analysis scope, which can distort the overall market picture. Additionally, due to the model's characteristic of placing higher weight on rankings themselves rather than individual app performance, top-tier apps' revenue curves tend to appear much more steadily flattened than their actual volatility, or when a new #1 app appears, it tends to be estimated with almost the exact same revenue as the previous #1 app. Furthermore, because the model targets only in-app purchases for estimation, currency sales through game companies' own websites, external payment systems, and cross-platform revenues linked via PC clients are not tracked, causing discrepancies between actual revenue and estimates.
Despite Obvious Limitations, Why Use the Platforms?
The key lies in utilizing one's own measured data as a calibration coefficient. First, calculate the error multiplier between your own game's actual revenue and the platform's estimate, and apply this multiplier to competitor estimates to retroactively calculate revenue.
Even if the estimation accuracy of platform data is only around 60%, companies can correct errors based on their internal data to reasonably infer competitors' performance and objectively evaluate their own market position. This will provide strong grounds for core decision-making such as setting marketing budgets, establishing goals for new projects, and attracting investments.