five

Potential Issues With Facebook's Advertising Algorithms: Part II

收藏
DataCite Commons2020-09-05 更新2024-07-25 收录
下载链接:
https://figshare.com/articles/dataset/Potential_Issues_With_Facebook_s_Advertising_Algorithms_Part_II/767337/1
下载链接
链接失效反馈
官方服务:
资源简介:
Attached you will find screenshots of various ads that we've released on Facebook. The conclusion: The software predicted a performance for all of the ads pictured of 0.01 - 0.03. Prior to having major issues with negative "SEO" tactics on our facebook page the software's predictions for advertising perforamnce was extremely accurate. After we began experiencing the troll problem, the advertisements began to perform in a very unpredicatble, sporadic, and oftentimes schizophrenic fashion. The same ad, with the swimsuit model, showed a sustained performance of 0.08 earlier in the day, and later in the evening, after a series of technial difficulties on Facebook's end, with our page, its perforamance shot down to 0.19. We took video of the ads from the moment they began to run until we paused them after the money started to deplete at an alarming rate, and what we found was shocking. In the first 30 minutes our page was growing at a rate of around 3 likes a second, which is about what we experienced when our ads were predicted to perform between 0.01 and 0.03 - and sustained those rates for days. Then suddenly, almost as if a tap was turned off, the likes stopped and the ad performance plunged to dismal numbers that do not accurately reflect the rate at which our likes were initially gained. We have paused all of our facebook ads and plan to thoroughly advertise this data. As stated in a previous article, which is linked below, we cannot say for sure that Facebook is using bot acconts because we do not have access to their internal systems, but this data highly suggests that they are using bot accounts, and that those bot accounts (as a direct result of not being human) are very senstive to negative SEO tactics (and we should also point out, conversely positive SEO tactics) becuse the human input is required to provide the Facebook bots with the necessary distribution of likes upon which to base the performance of the ads. Without this, assuming Facebook is using fake bot accounts (as accused in reference 1) there would be no way for Facebook's army of drone accounts to accurately decide whether or not content is "cool." Imagine a situation where an army of fake Facebook bot accounts liked a blank picture with nonsense text. This might also explain Facebook's draconia advertising restrictions regarding text on the face of ads, etc. Again, as stated before, we cannot prove any of this without access to Facebook's internal list of accounts for investigation, but this is what our data suggests. A video of the rate of growth will be posted later. Reference: 1. http://www.forbes.com/sites/davidthier/2012/08/01/facebook-investigating-claims-that-80-of-ad-clicks-come-from-bots/ 2. Previous figshare article: http://figshare.com/articles/Potential_Issues_with_FB_Advertising_Algorithms_/767331
提供机构:
figshare
创建时间:
2016-01-18
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作