Click Fraud Forensics Dean Qudah Pace University DPS 2010.

9
Click Fraud Click Fraud Forensics Forensics Dean Qudah Dean Qudah Pace University Pace University DPS 2010 DPS 2010

Transcript of Click Fraud Forensics Dean Qudah Pace University DPS 2010.

Click Fraud Click Fraud Forensics Forensics

Dean Qudah Dean Qudah Pace UniversityPace University

DPS 2010DPS 2010

What Is Click Fraud?What Is Click Fraud?

Click fraud, the intentional clicking on PPC Click fraud, the intentional clicking on PPC advertisements, where the perpetrator has advertisements, where the perpetrator has no intention of buying the products or no intention of buying the products or services advertisedservices advertised

Elements of PPC Elements of PPC AdvertisingAdvertising

Click Fraud Categories Click Fraud Categories

Clicking on competitors occurs when a Clicking on competitors occurs when a company purposely clicks on a competitor company purposely clicks on a competitor so as to cost them money, use up their so as to cost them money, use up their daily budgets, and force them off the daily budgets, and force them off the auction. auction.

Network fraud occurs when Website Network fraud occurs when Website owners click on their own banner owners click on their own banner advertisements in order to generate advertisements in order to generate revenue from the search engine who is revenue from the search engine who is serving the banner advertisement.serving the banner advertisement.

How Big Is It?How Big Is It?

Click Forensics, a US firm that audits Click Forensics, a US firm that audits Internet traffic, reported that 17.1 percent Internet traffic, reported that 17.1 percent of clicks on online advertising were frauds of clicks on online advertising were frauds evidently intended solely to drive up bills evidently intended solely to drive up bills for businesses paying "per click.”for businesses paying "per click.”

Networks of hacked computers referred to Networks of hacked computers referred to as "botnets" are said to be responsible for as "botnets" are said to be responsible for nearly a third of the click fraud in final three nearly a third of the click fraud in final three months of 2008.months of 2008.

Can We trust Search Can We trust Search Engines?Engines?

Google has trivialized click fraud and Google has trivialized click fraud and mischaracterized it as a minor problem.mischaracterized it as a minor problem.

After Many Law suits and hundreds of After Many Law suits and hundreds of reports from different agencies, Google reports from different agencies, Google Admitted the problem and start Admitted the problem and start cooperating. cooperating.

After all Click Fraud means more profit for After all Click Fraud means more profit for search engines.search engines.

Click Fraud DetectionClick Fraud Detection

There are many ways to detect and minimize click There are many ways to detect and minimize click fraud;fraud;

• Monitoring the Monitoring the Click Through Rate ( CTR), and Click Through Rate ( CTR), and make sure it is within the norms.make sure it is within the norms.

• The Data Analysis Approach. Since privacy is a The Data Analysis Approach. Since privacy is a major obstacle in detecting fraud, we have no major obstacle in detecting fraud, we have no choice but to use data analysis approach by choice but to use data analysis approach by analyzing the temporary surfers’ Identification analyzing the temporary surfers’ Identification data such as IP address and Cookies. data such as IP address and Cookies.

Example of Statistical Example of Statistical MethodsMethods

Statistical data analysis can be done by Statistical data analysis can be done by storing the entire traffic data in a database storing the entire traffic data in a database and periodically executing aggregate SQL and periodically executing aggregate SQL queries to identify outliers that would be queries to identify outliers that would be candidate for fraudulent behavior.candidate for fraudulent behavior.

This method has a challenging scale This method has a challenging scale problem here. Since an average-sized problem here. Since an average-sized commissioner receives around 70M records commissioner receives around 70M records per hour, storing the traffic and running the per hour, storing the traffic and running the query could be very expensive. query could be very expensive.

Mission un Mission un accomplishedaccomplished

So far there is no good solution for this So far there is no good solution for this problem.problem.

Fraudsters become more and more Fraudsters become more and more advanced and use automated tools and advanced and use automated tools and software. software.

Many research papers and PhD dissertation Many research papers and PhD dissertation tried to come up with solutions.tried to come up with solutions.