Web Science/Part2: Emerging Web Properties/Web Search Ecosystem/Understanding the Problems with Click Fraud/script

As we have seen the online advertising market is very big and most ads are nowadays sold with a CPC model. Since we have many different stakeholders interacting with each other some of the stakeholders see opportunities for fraudulent behavior. In the following we will discuss the types of click fraud and who is benefiting and suffering from it. We will also have a brief look at the technical details about how detection of click fraud is possible.

As a publisher displaying CPC ads from some ad agency you could have the idea that your revenue will be higher if you click the advertisement yourself even though you are not interested in the ad. This is the first source of click fraud. Obviously the advertiser does not want to receive those clicks since there is a very low chance in converting a person who is not interested in the advertisement but just in the click of the banner for a personal profit. Obviously the ad agency is not interested in this kind of click fraud. Even though they also earn some margin of the CPC and could be happy about more clicks they know that the advertiser will not be satisfied with the traffic driven from this campaign and on the long term consider changing the ad provider. This type of fraud is often prohibited in the terms of service between the ad agency and the publisher. So if he is being detected clicking on his own advertisement he will have to pay high fines.

A second source of fraud could be coming from the competitor of the advertiser. Even though he has no interest in bringing money to the publisher and ad agency he might want to lower the success of the campaign of the advertiser. By clicking on the ads but bouncing from the site he could soak a lot of money from the advertiser who will be charged for each click and taking his chances away of getting to know new customers. This type of click fraud is legally not so well defined as the first setting. The competitor has no contract with the advertiser or the agency. Even though we clearly realize that such a behavior should not be acceptable there cannot be any terms of services which would prohibit the competitor to do something like this. So this situation would need a regulation from the government. On the other site there is also nothing wrong with surfing the internet and clicking links. Even if there was a regulation for this case, it would still be very hard to distinguish the fraudulent clicks from real clicks that just bounced away.

There are many other forms of click fraud and reasons that make people believe they should commit click fraud. There are even some settings in which the publisher is the victim of the fraudulent attack. What is important for us is to understand that defining what exactly click fraud is and how to detect it might be very difficult. For example one could argue if one particular IP address drives most clicks that this could be just a single user producing fraudulent clicks. But maybe this IP address as operating as a NAT for the visitors of a large event and all the clicks are coming from real people. It will just be impossible to know for the ad agency. On the other hand there exists a market where you can buy click fraud. People offering click fraud as a service will most likely operate bot nets or have people in many regions sitting in order to simulate real traffic. Such a system is often called a click farm. This could also be operated from countries where labor is cheap or in combination with crowd sourcing platforms like amazon mechanical turk (which is prohibited: https://www.mturk.com/mturk/help?helpPage=policies ). Ad agencies of course try to fight click fraud by analyzing their log files. This can be even more effective by using more tracking and cookies. This on the other side will have an impact to the advertisers, publishers and consumers privacy.