{
  "id": 55054,
  "title": "Business Context of Click Fraud - Chinese Click Farm - V2",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/55054",
  "author_name": "",
  "post_date": "2018-04-21T15:18:57.082030700Z",
  "votes": 26,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hello all, I have already written a <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/54765\">post</a> covering some of the basic reality of \"chinese click farm\" and seems that it is rather well taken :)   So I would like to follow-up with some in-depth information, and I hope what is revealed here will be helpful for both understanding this problem, and creating useful features. Most of the information I provide here is taken from this <a href=\"http://www.shejipi.com/158829.html\">post</a> - took me sometime to do the translation.</p>\n\n<p>TL;RD is the following: </p>\n\n<ol>\n<li>App creators create app, and need to promote such app to top of App LB to generate visability/income/business</li>\n<li>App creator pay ad channel to advertise their app</li>\n<li>Ad channel uses various methods to promote app, some via honest effort and some (mostly?) via less honest approach - there comes fraudulent click &amp; download</li>\n<li>Many App creators are knowling employing these ad channel, since traditional promotional cycle can take years and are obviously too long for small/start-up business, while viral apps like Angry Bird are extremely rare. </li>\n</ol>\n\n<p>A bit more details: , first of all, different kinds of click farming </p>\n\n<ul>\n<li>Human based click farming </li>\n<li>Machine based click farming</li>\n</ul>\n\n<p><strong>Human based click farming</strong>, there are at least two kinds: 1) manually clicking in a workshop style setting, as you see in the previous post - I won't elaborate more here,  2) social based click farming, and this one can be explained as the following:</p>\n\n<p>There exist \"click farming platforms\",  these platforms provide  click farming apps to  up to millions of users.   See below for a screenshot of such an app.  for these user, each time they download a specific app on such platform they will get paid about 2 RMB, and each day they generate income around several to tens of RMB. In additional to these, each user can bring in “followers” - such followers are new user that are brought onto the platform by existing users. Existing users get additional income when their follower download apps. </p>\n\n<p><img src=\"https://snag.gy/Oi1QT2.jpg\" alt=\"enter image description here\"></p>\n\n<p>These ad channel charge app makers/creators 3-4 RMB for each of new user (i.e. new download) - therefore , set aside the amount they pay the users, and considering that they have millions of users, these ad channel  generate millions of RMB per day - and this is relatively conservative estimation. </p>\n\n<p><strong>Machie based click farming</strong>, there are different level of machine automation for this. You can have <em>complete simulation</em> - i.e. by reverse engineering App Store, in a specific time, simulate user search, download, installation events -  in such case you don’t need real mobiles, you need a comptuer + a hacker um… a “professional”. Needlessly to say this is  technically more difficult to achieve.</p>\n\n<p>a less automated kind, is such as the one you see in the previous post: i.e. workshop with many mobiles, each of this mobile is wired up to infrastructure that run clicking/downloading script, as well as “one click refactorying” to change mobile identity - <strong>one mobile can assume several hundreds identities in its lifecycle this way.</strong>  I think this point is interesting in the context of this competition.</p>\n\n<p>Now, <strong>pricing models</strong>: \npricing model is oriented toward achieving top ranking in various LB - One shall know the imporatnce of LB by the mere fact that you are reading this - Free App LB, LB for specific category (gaming, tool, entertainment, etc) </p>\n\n<p>See the following image for such an price catelouge for Apple AppStore:\n <img src=\"https://snag.gy/YKEjnP.jpg\" alt=\"enter image description here\"></p>\n\n<p>Some translation for you to understand the above table:</p>\n\n<p><img src=\"https://snag.gy/oavzJ8.jpg\" alt=\"enter image description here\"></p>\n\n<p>so, price for pushing an app to top 10 free app LB for one single day is roughly $11117 USD, nice eh? Warmly welcome you to a China's very VIRTUAL digital market :)</p>",
  "messages": [
    {
      "id": "317437",
      "postDate": "04/21/2018 15:18:57",
      "content": "<p>Hello all, I have already written a <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/54765\">post</a> covering some of the basic reality of \"chinese click farm\" and seems that it is rather well taken :)   So I would like to follow-up with some in-depth information, and I hope what is revealed here will be helpful for both understanding this problem, and creating useful features. Most of the information I provide here is taken from this <a href=\"http://www.shejipi.com/158829.html\">post</a> - took me sometime to do the translation.</p>\n\n<p>TL;RD is the following: </p>\n\n<ol>\n<li>App creators create app, and need to promote such app to top of App LB to generate visability/income/business</li>\n<li>App creator pay ad channel to advertise their app</li>\n<li>Ad channel uses various methods to promote app, some via honest effort and some (mostly?) via less honest approach - there comes fraudulent click &amp; download</li>\n<li>Many App creators are knowling employing these ad channel, since traditional promotional cycle can take years and are obviously too long for small/start-up business, while viral apps like Angry Bird are extremely rare. </li>\n</ol>\n\n<p>A bit more details: , first of all, different kinds of click farming </p>\n\n<ul>\n<li>Human based click farming </li>\n<li>Machine based click farming</li>\n</ul>\n\n<p><strong>Human based click farming</strong>, there are at least two kinds: 1) manually clicking in a workshop style setting, as you see in the previous post - I won't elaborate more here,  2) social based click farming, and this one can be explained as the following:</p>\n\n<p>There exist \"click farming platforms\",  these platforms provide  click farming apps to  up to millions of users.   See below for a screenshot of such an app.  for these user, each time they download a specific app on such platform they will get paid about 2 RMB, and each day they generate income around several to tens of RMB. In additional to these, each user can bring in “followers” - such followers are new user that are brought onto the platform by existing users. Existing users get additional income when their follower download apps. </p>\n\n<p><img src=\"https://snag.gy/Oi1QT2.jpg\" alt=\"enter image description here\"></p>\n\n<p>These ad channel charge app makers/creators 3-4 RMB for each of new user (i.e. new download) - therefore , set aside the amount they pay the users, and considering that they have millions of users, these ad channel  generate millions of RMB per day - and this is relatively conservative estimation. </p>\n\n<p><strong>Machie based click farming</strong>, there are different level of machine automation for this. You can have <em>complete simulation</em> - i.e. by reverse engineering App Store, in a specific time, simulate user search, download, installation events -  in such case you don’t need real mobiles, you need a comptuer + a hacker um… a “professional”. Needlessly to say this is  technically more difficult to achieve.</p>\n\n<p>a less automated kind, is such as the one you see in the previous post: i.e. workshop with many mobiles, each of this mobile is wired up to infrastructure that run clicking/downloading script, as well as “one click refactorying” to change mobile identity - <strong>one mobile can assume several hundreds identities in its lifecycle this way.</strong>  I think this point is interesting in the context of this competition.</p>\n\n<p>Now, <strong>pricing models</strong>: \npricing model is oriented toward achieving top ranking in various LB - One shall know the imporatnce of LB by the mere fact that you are reading this - Free App LB, LB for specific category (gaming, tool, entertainment, etc) </p>\n\n<p>See the following image for such an price catelouge for Apple AppStore:\n <img src=\"https://snag.gy/YKEjnP.jpg\" alt=\"enter image description here\"></p>\n\n<p>Some translation for you to understand the above table:</p>\n\n<p><img src=\"https://snag.gy/oavzJ8.jpg\" alt=\"enter image description here\"></p>\n\n<p>so, price for pushing an app to top 10 free app LB for one single day is roughly $11117 USD, nice eh? Warmly welcome you to a China's very VIRTUAL digital market :)</p>",
      "rawMarkdown": "Hello all, I have already written a [post][1] covering some of the basic reality of \"chinese click farm\" and seems that it is rather well taken :)   So I would like to follow-up with some in-depth information, and I hope what is revealed here will be helpful for both understanding this problem, and creating useful features. Most of the information I provide here is taken from this [post][2] - took me sometime to do the translation.\n\nTL;RD is the following: \n\n 1. App creators create app, and need to promote such app to top of App LB to generate visability/income/business\n 2. App creator pay ad channel to advertise their app\n 3. Ad channel uses various methods to promote app, some via honest effort and some (mostly?) via less honest approach - there comes fraudulent click &amp; download\n 4. Many App creators are knowling employing these ad channel, since traditional promotional cycle can take years and are obviously too long for small/start-up business, while viral apps like Angry Bird are extremely rare. \n\nA bit more details: , first of all, different kinds of click farming \n\n - Human based click farming \n - Machine based click farming\n\n**Human based click farming**, there are at least two kinds: 1) manually clicking in a workshop style setting, as you see in the previous post - I won't elaborate more here,  2) social based click farming, and this one can be explained as the following:\n\nThere exist \"click farming platforms\",  these platforms provide  click farming apps to  up to millions of users.   See below for a screenshot of such an app.  for these user, each time they download a specific app on such platform they will get paid about 2 RMB, and each day they generate income around several to tens of RMB. In additional to these, each user can bring in “followers” - such followers are new user that are brought onto the platform by existing users. Existing users get additional income when their follower download apps. \n\n![enter image description here][3]\n\nThese ad channel charge app makers/creators 3-4 RMB for each of new user (i.e. new download) - therefore , set aside the amount they pay the users, and considering that they have millions of users, these ad channel  generate millions of RMB per day - and this is relatively conservative estimation. \n\n\n**Machie based click farming**, there are different level of machine automation for this. You can have *complete simulation* - i.e. by reverse engineering App Store, in a specific time, simulate user search, download, installation events -  in such case you don’t need real mobiles, you need a comptuer + a hacker um… a “professional”. Needlessly to say this is  technically more difficult to achieve.\n\na less automated kind, is such as the one you see in the previous post: i.e. workshop with many mobiles, each of this mobile is wired up to infrastructure that run clicking/downloading script, as well as “one click refactorying” to change mobile identity - **one mobile can assume several hundreds identities in its lifecycle this way.**  I think this point is interesting in the context of this competition.\n\n\nNow, **pricing models**: \npricing model is oriented toward achieving top ranking in various LB - One shall know the imporatnce of LB by the mere fact that you are reading this - Free App LB, LB for specific category (gaming, tool, entertainment, etc) \n\nSee the following image for such an price catelouge for Apple AppStore:\n ![enter image description here][4]\n\nSome translation for you to understand the above table:\n\n![enter image description here][5]\n\nso, price for pushing an app to top 10 free app LB for one single day is roughly $11117 USD, nice eh? Warmly welcome you to a China's very VIRTUAL digital market :)\n\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/54765\n  [2]: http://www.shejipi.com/158829.html\n  [3]: https://snag.gy/Oi1QT2.jpg\n  [4]: https://snag.gy/YKEjnP.jpg\n  [5]: https://snag.gy/oavzJ8.jpg",
      "votes": null
    },
    {
      "id": "317640",
      "postDate": "04/22/2018 04:42:54",
      "content": "<p>Some information here too</p>\n\n<p><a href=\"https://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0\">https://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0</a></p>",
      "rawMarkdown": "Some information here too\n\nhttps://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0",
      "votes": null
    },
    {
      "id": "318681",
      "postDate": "04/24/2018 09:06:46",
      "content": "<p>Thanks for providing some background info, just from the description given by TalkingData, it doesn't become particularly clear what the exact context of this data is. Do you think that this a problem that is especially prevalent in China, and if so, why would that be the case?</p>",
      "rawMarkdown": "Thanks for providing some background info, just from the description given by TalkingData, it doesn't become particularly clear what the exact context of this data is. Do you think that this a problem that is especially prevalent in China, and if so, why would that be the case?",
      "votes": null
    },
    {
      "id": "318688",
      "postDate": "04/24/2018 09:15:49",
      "content": "<p>Short answer, no it is not specific to China, check out the link shared <a href=\"/hengck23\">@hengck23</a> above</p>",
      "rawMarkdown": "Short answer, no it is not specific to China, check out the link shared @hengck23 above",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 317640,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "04/22/2018 04:42:54",
      "content": "<p>Some information here too</p>\n\n<p><a href=\"https://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0\">https://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 318681,
      "author_name": "baseratezero",
      "author_url": "",
      "post_date": "04/24/2018 09:06:46",
      "content": "<p>Thanks for providing some background info, just from the description given by TalkingData, it doesn't become particularly clear what the exact context of this data is. Do you think that this a problem that is especially prevalent in China, and if so, why would that be the case?</p>",
      "votes": null,
      "replies": [
        {
          "id": 318688,
          "author_name": "yifanxie",
          "author_url": "",
          "post_date": "04/24/2018 09:15:49",
          "content": "<p>Short answer, no it is not specific to China, check out the link shared <a href=\"/hengck23\">@hengck23</a> above</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "317437": "Hello all, I have already written a [post][1] covering some of the basic reality of \"chinese click farm\" and seems that it is rather well taken :)   So I would like to follow-up with some in-depth information, and I hope what is revealed here will be helpful for both understanding this problem, and creating useful features. Most of the information I provide here is taken from this [post][2] - took me sometime to do the translation.\n\nTL;RD is the following: \n\n 1. App creators create app, and need to promote such app to top of App LB to generate visability/income/business\n 2. App creator pay ad channel to advertise their app\n 3. Ad channel uses various methods to promote app, some via honest effort and some (mostly?) via less honest approach - there comes fraudulent click &amp; download\n 4. Many App creators are knowling employing these ad channel, since traditional promotional cycle can take years and are obviously too long for small/start-up business, while viral apps like Angry Bird are extremely rare. \n\nA bit more details: , first of all, different kinds of click farming \n\n - Human based click farming \n - Machine based click farming\n\n**Human based click farming**, there are at least two kinds: 1) manually clicking in a workshop style setting, as you see in the previous post - I won't elaborate more here,  2) social based click farming, and this one can be explained as the following:\n\nThere exist \"click farming platforms\",  these platforms provide  click farming apps to  up to millions of users.   See below for a screenshot of such an app.  for these user, each time they download a specific app on such platform they will get paid about 2 RMB, and each day they generate income around several to tens of RMB. In additional to these, each user can bring in “followers” - such followers are new user that are brought onto the platform by existing users. Existing users get additional income when their follower download apps. \n\n![enter image description here][3]\n\nThese ad channel charge app makers/creators 3-4 RMB for each of new user (i.e. new download) - therefore , set aside the amount they pay the users, and considering that they have millions of users, these ad channel  generate millions of RMB per day - and this is relatively conservative estimation. \n\n\n**Machie based click farming**, there are different level of machine automation for this. You can have *complete simulation* - i.e. by reverse engineering App Store, in a specific time, simulate user search, download, installation events -  in such case you don’t need real mobiles, you need a comptuer + a hacker um… a “professional”. Needlessly to say this is  technically more difficult to achieve.\n\na less automated kind, is such as the one you see in the previous post: i.e. workshop with many mobiles, each of this mobile is wired up to infrastructure that run clicking/downloading script, as well as “one click refactorying” to change mobile identity - **one mobile can assume several hundreds identities in its lifecycle this way.**  I think this point is interesting in the context of this competition.\n\n\nNow, **pricing models**: \npricing model is oriented toward achieving top ranking in various LB - One shall know the imporatnce of LB by the mere fact that you are reading this - Free App LB, LB for specific category (gaming, tool, entertainment, etc) \n\nSee the following image for such an price catelouge for Apple AppStore:\n ![enter image description here][4]\n\nSome translation for you to understand the above table:\n\n![enter image description here][5]\n\nso, price for pushing an app to top 10 free app LB for one single day is roughly $11117 USD, nice eh? Warmly welcome you to a China's very VIRTUAL digital market :)\n\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/54765\n  [2]: http://www.shejipi.com/158829.html\n  [3]: https://snag.gy/Oi1QT2.jpg\n  [4]: https://snag.gy/YKEjnP.jpg\n  [5]: https://snag.gy/oavzJ8.jpg",
    "317640": "Some information here too\n\nhttps://medium.com/swlh/fraud-traffic-the-dark-side-of-mobile-digital-advertising-ead808e054d0",
    "318681": "Thanks for providing some background info, just from the description given by TalkingData, it doesn't become particularly clear what the exact context of this data is. Do you think that this a problem that is especially prevalent in China, and if so, why would that be the case?",
    "318688": "Short answer, no it is not specific to China, check out the link shared @hengck23 above"
  },
  "source": "meta"
}