Ad Network Click Fraud

[164]

I read on google adsense support, the company claims it uses 200+ metrics to track real traffic from fraudulent ones. Are they just lies to make advertisers feel safe, or they are real in fact. If so what are some of those metrics or the common ones. Also do all ad networks fingerprint browsers
p.s. I am also trying to build an ad bot network to imitate real traffic. It's based on selenium

[33]

According to my knowledge the number of visits from the same ip or list of ip is one and the way the visitor act and use the website before clicking an ad that behaviour on the website is by it self a set of different metrics taken together to see if the visitor is real. Of course some people just visit a link and don't do anything and go out or click the link if it is done once it is a real traffic if it done multiple times then no its not. If a huge list of IPs is used maybe it will be seen as real traffic at first and then not. If the program you make can visit the websites and do different different action where the action is not monotonic and mirror a real used and then click and ad or such it is seen as real traffic but if that is done a lot then no. People (humans and not machines) don't click hundreds of ads everyday and don't visit the same website hundreds of times a day or in a heuristical manner. A machine or an Ip is taken as 1 human. If you are talking about different websites using the same ad provider then the provider will know that you used a different website to click ad and it is all taken into consideration.

Just think of how you as a normal human use the internet and how you act in the presence of ads and how much you click. And think about the amount of traffic you generate everyday.

If you want to generate a huge amount of traffic that seems real then you need a huge amount of machines and those machines must be located in different part of the globe and not in the same city(we forgot to talk about location which is connected to ip over the internet a little). 

So using a machine to automate a real person action to the limit for traffic is the only way. And for more traffic you need machines to simulate how people around the world use the internet.

Machine Learning and Deep Learning can analyse the data to know the fraud in case it is not truly a person accept if a machine can emulate a person if you use the data of how people use the internet and train a machine learning model to do the job it would be the only way in my opinion.

[164]

According to my knowledge the number of visits from the same ip or list of ip is one and the way the visitor act and use the website before clicking an ad that behaviour on the website is by it self a set of different metrics taken together to see if the visitor is real. Of course some people just visit a link and don't do anything and go out or click the link if it is done once it is a real traffic if it done multiple times then no its not. If a huge list of IPs is used maybe it will be seen as real traffic at first and then not. If the program you make can visit the websites and do different different action where the action is not monotonic and mirror a real used and then click and ad or such it is seen as real traffic but if that is done a lot then no. People (humans and not machines) don't click hundreds of ads everyday and don't visit the same website hundreds of times a day or in a heuristical manner. A machine or an Ip is taken as 1 human. If you are talking about different websites using the same ad provider then the provider will know that you used a different website to click ad and it is all taken into consideration.

Just think of how you as a normal human use the internet and how you act in the presence of ads and how much you click. And think about the amount of traffic you generate everyday.

If you want to generate a huge amount of traffic that seems real then you need a huge amount of machines and those machines must be located in different part of the globe and not in the same city(we forgot to talk about location which is connected to ip over the internet a little). 

So using a machine to automate a real person action to the limit for traffic is the only way. And for more traffic you need machines to simulate how people around the world use the internet.

Machine Learning and Deep Learning can analyse the data to know the fraud in case it is not truly a person accept if a machine can emulate a person if you use the data of how people use the internet and train a machine learning model to do the job it would be the only way in my opinion.

Thanks a lot, thou i plan on using nordvpn, ipvanish and expressvpn via openvpn to swap out ips continously for each unique visitor sessions on 60 different machine instances open on the cloud while using the scroll, arrowkeys and scroll bar to interact with the page and the use of bezier curves to control mouse movements while maintaining the average page session time to 2 minutes and 8% ctr for all traffic. I do think it will succeed because i've come to the conclusion that ad networks don't go to the extreme to protect advertisers money, because of their selfish interest to increase their revenue. I hope to give you a feedback on this endeavor

[113]

Thanks a lot, thou i plan on using nordvpn, ipvanish and expressvpn via openvpn to swap out ips continously for each unique visitor sessions on 60 different machine instances open on the cloud while using the scroll, arrowkeys and scroll bar to interact with the page and the use of bezier curves to control mouse movements while maintaining the average page session time to 2 minutes and 8% ctr for all traffic. I do think it will succeed because i've come to the conclusion that ad networks don't go to the extreme to protect advertisers money, because of their selfish interest to increase their revenue. I hope to give you a feedback on this endeavor

You will likely need residential proxies to create these tools as VPNs do not have many IPs available.