What is hidden under the ranking algorithm in the Apple App Store? Habra Quest
Introduction
When talking between people it comes to mobile applications, one often hears about astronomical amounts earned by one or another world-famous developer or a huge number of downloads of an application. The media now and then report on the launch of the ISS plush pigs from Angry Birds, and in the US, Zuckerberg bought Instagram for $ 1,000,000,000.
Many people like to talk about mobile apps. This is a modern, interesting, finally, fashionable theme and really deserves attention.
')
Soon after we started doing the Apps4All project in the summer of 2011, I began to ask myself how many times they downloaded or how much this or that application earns. Communicating with other developers and representatives of venture funds and corporations, I also noticed that they are often interested in data about the success of mobile applications.
This data was not easy to find ...
On the Internet, we have with you all kinds of counters and ratings. Finally, there are services like www.comscore.com and www.alexa.com , with the help of which you can certainly estimate the attendance of a particular site.
Also on the Internet, TV, Radio and some other media there are rating agencies, such as TNS Gallup, through which the advertising market operates. (They measure the audience of channels, depending on which advertising pricing occurs)
The press uses the circulation commissions, which confirm the size of the circulation of publications.
And what about mobile apps?
Access closed.
What is the reality of the modern economy of mobile applications?
• A closed market, 90 percent of which is controlled by 3 corporations, among which there is cartel collusion. (How else can one explain that the commission of all major app stores is exactly 30 percent? Why has no one done 29 to attract more developers?)
Full control at all stages of the distribution of applications on the one hand ensures the quality control and stability of the ecosystem, on the other hand makes it closed and opaque.
In reality, the economy of mobile applications is growing at a tremendous pace and everyone understands this. At the moment, more than 1 million applications have been published in aggregate in three stores and there is no doubt that this number will double in the coming years. Already, to succeed in the market of mobile applications, it is not enough just to “weaken the application”; only truly high-quality projects survive and the time of iFart-type applications (simulator of gas formation in the rectum) is irretrievably gone.
Lack of data
Developers, investors, analysts, advertising agents, freelancers, users need analytics of mobile applications.
To be successful, you need to look around you, see what other market participants are doing and plan.
The specialist should ask questions:
• How are my competitors doing?
• Where is the free market niche now?
• How did the marketing campaign affect downloads of this or that app? Can we use this tool too?
• Which monetization model to choose for my platform or application category?
And so on.
Unfortunately, Apple, Google and Microsoft do not disclose to us (their platform users and developers) any data about application downloads or store statistics, in any form or on a regular basis. All that we can be content with is rare, non-specific press releases and reports from third-party analysts.
Investigation
In the spring of 2012 a small start-up was launched in Berlin, consisting of 3 people, two young people and a girl. The main mission of entrepreneurs was to create a single ranking algorithm for all mobile applications, like Google PageRank.
For reference, Google PageRank on Wiki data:
PageRank is one of the reference ranking algorithms. The algorithm is applied to a collection of documents linked by hyperlinks (such as web pages from the world wide web), and assigns each of them a certain numerical value, which measures its “importance” or “credibility” among other documents. Generally speaking, the algorithm can be applied not only to web pages, but also to any set of objects linked by reciprocal links, that is, to any graph.
Not so long ago, at the end of July, they received funding and began to glow actively in the US at techcrunch and other specialized sites. This service predicts how many times this or that application downloaded. Really curious.
You can familiarize yourself with this service here - www.xyologic.com
But our team was tormented by the question: “Did they really solve the App Store ranking algorithm?!”
The most comprehensive answer on the Internet to this question was found on Quora.
www.quora.com/What-is-the-algorithm-behind-the-App-Store-rankings
I give below a partial translation:
Brandon Smietana, Founder of Symbolic Analytics
The algorithm has several components, most of which are not known. We know that the algorithm uses a weighted average number of app sales. Basically, only the last 4 days of sales are counted. You can determine weights for a 4-day weighted average based on your sales rating data and a least-squares analysis. In order to improve measurement accuracy, you must normalize your sales to total sales in the App Store.
Applications are ranked not by sold units, but by aggregated revenue. This is due to the fact that large game makers such as Electronic Arts lobbied Apple for this system. They do not feel comfortable when amateur developers bypass them in the rating, and this prevents their applications with a price of $ 10 from competing with applications of $ 1. Therefore, Apple changed the ranking system so that in the first place were the applications that generate more profits, and not those that download more. There is one more explanation for this - those applications that bring great profits to Apple itself occupy the first positions.
Another opinion:
Chris Lee
I would like to point out that in the App Store only income is used. There are 3 types of ratings: Top Paid, Top Free and Top Cash. These categories fall under the iPhone and / or iPad App Store. Each category of applications (for example, business, performance, games, etc.) also has a sub-rating.
Rating of the highest grossing uses the income from the sale of applications. The more income the application generates, the higher the rank in the ranking of the highest grossing. For example, a $ 10 app sold 10 times ($ 100) and a $ 1 app sold 15 times. ($ 15) Then, the $ 10 app will have a higher rating in this table.
Top Free and Top Paid, on the other hand, are based on the number of downloads. The best free speaks for itself. The higher the load, the higher the rank. In the Top Paid rating, the $ 1 app will have a higher rating in the above scenario.
And one more hypothesis:
www.readwriteweb.com/start/2010/02/iphone-appstore-ranking-algorithm.php
Conclusion
Our team has been working on the disclosure of the ranking formula for several months. In this article I shared my experience in order to find in Habra a community of people who, like me, are interested in this issue.
We would like to initiate brainstorming in the Habr community in order to work together to solve this riddle.
Creating a high-quality and open tool for projection of downloads and application revenue will make the global mobile application market more mature and transparent.
When talking between people it comes to mobile applications, one often hears about astronomical amounts earned by one or another world-famous developer or a huge number of downloads of an application. The media now and then report on the launch of the ISS plush pigs from Angry Birds, and in the US, Zuckerberg bought Instagram for $ 1,000,000,000.
Many people like to talk about mobile apps. This is a modern, interesting, finally, fashionable theme and really deserves attention.
')
Soon after we started doing the Apps4All project in the summer of 2011, I began to ask myself how many times they downloaded or how much this or that application earns. Communicating with other developers and representatives of venture funds and corporations, I also noticed that they are often interested in data about the success of mobile applications.
This data was not easy to find ...
On the Internet, we have with you all kinds of counters and ratings. Finally, there are services like www.comscore.com and www.alexa.com , with the help of which you can certainly estimate the attendance of a particular site.
Also on the Internet, TV, Radio and some other media there are rating agencies, such as TNS Gallup, through which the advertising market operates. (They measure the audience of channels, depending on which advertising pricing occurs)
The press uses the circulation commissions, which confirm the size of the circulation of publications.
And what about mobile apps?
Access closed.
What is the reality of the modern economy of mobile applications?
• A closed market, 90 percent of which is controlled by 3 corporations, among which there is cartel collusion. (How else can one explain that the commission of all major app stores is exactly 30 percent? Why has no one done 29 to attract more developers?)
Full control at all stages of the distribution of applications on the one hand ensures the quality control and stability of the ecosystem, on the other hand makes it closed and opaque.
In reality, the economy of mobile applications is growing at a tremendous pace and everyone understands this. At the moment, more than 1 million applications have been published in aggregate in three stores and there is no doubt that this number will double in the coming years. Already, to succeed in the market of mobile applications, it is not enough just to “weaken the application”; only truly high-quality projects survive and the time of iFart-type applications (simulator of gas formation in the rectum) is irretrievably gone.
Lack of data
Developers, investors, analysts, advertising agents, freelancers, users need analytics of mobile applications.
To be successful, you need to look around you, see what other market participants are doing and plan.
The specialist should ask questions:
• How are my competitors doing?
• Where is the free market niche now?
• How did the marketing campaign affect downloads of this or that app? Can we use this tool too?
• Which monetization model to choose for my platform or application category?
And so on.
Unfortunately, Apple, Google and Microsoft do not disclose to us (their platform users and developers) any data about application downloads or store statistics, in any form or on a regular basis. All that we can be content with is rare, non-specific press releases and reports from third-party analysts.
Investigation
In the spring of 2012 a small start-up was launched in Berlin, consisting of 3 people, two young people and a girl. The main mission of entrepreneurs was to create a single ranking algorithm for all mobile applications, like Google PageRank.
For reference, Google PageRank on Wiki data:
PageRank is one of the reference ranking algorithms. The algorithm is applied to a collection of documents linked by hyperlinks (such as web pages from the world wide web), and assigns each of them a certain numerical value, which measures its “importance” or “credibility” among other documents. Generally speaking, the algorithm can be applied not only to web pages, but also to any set of objects linked by reciprocal links, that is, to any graph.
Not so long ago, at the end of July, they received funding and began to glow actively in the US at techcrunch and other specialized sites. This service predicts how many times this or that application downloaded. Really curious.
You can familiarize yourself with this service here - www.xyologic.com
But our team was tormented by the question: “Did they really solve the App Store ranking algorithm?!”
The most comprehensive answer on the Internet to this question was found on Quora.
www.quora.com/What-is-the-algorithm-behind-the-App-Store-rankings
I give below a partial translation:
Brandon Smietana, Founder of Symbolic Analytics
The algorithm has several components, most of which are not known. We know that the algorithm uses a weighted average number of app sales. Basically, only the last 4 days of sales are counted. You can determine weights for a 4-day weighted average based on your sales rating data and a least-squares analysis. In order to improve measurement accuracy, you must normalize your sales to total sales in the App Store.
Applications are ranked not by sold units, but by aggregated revenue. This is due to the fact that large game makers such as Electronic Arts lobbied Apple for this system. They do not feel comfortable when amateur developers bypass them in the rating, and this prevents their applications with a price of $ 10 from competing with applications of $ 1. Therefore, Apple changed the ranking system so that in the first place were the applications that generate more profits, and not those that download more. There is one more explanation for this - those applications that bring great profits to Apple itself occupy the first positions.
Another opinion:
Chris Lee
I would like to point out that in the App Store only income is used. There are 3 types of ratings: Top Paid, Top Free and Top Cash. These categories fall under the iPhone and / or iPad App Store. Each category of applications (for example, business, performance, games, etc.) also has a sub-rating.
Rating of the highest grossing uses the income from the sale of applications. The more income the application generates, the higher the rank in the ranking of the highest grossing. For example, a $ 10 app sold 10 times ($ 100) and a $ 1 app sold 15 times. ($ 15) Then, the $ 10 app will have a higher rating in this table.
Top Free and Top Paid, on the other hand, are based on the number of downloads. The best free speaks for itself. The higher the load, the higher the rank. In the Top Paid rating, the $ 1 app will have a higher rating in the above scenario.
And one more hypothesis:
www.readwriteweb.com/start/2010/02/iphone-appstore-ranking-algorithm.php
Conclusion
Our team has been working on the disclosure of the ranking formula for several months. In this article I shared my experience in order to find in Habra a community of people who, like me, are interested in this issue.
We would like to initiate brainstorming in the Habr community in order to work together to solve this riddle.
Creating a high-quality and open tool for projection of downloads and application revenue will make the global mobile application market more mature and transparent.
Source: https://habr.com/ru/post/150702/
All Articles