The Lane’s Gifts vs. Google Report part 2Google’s Approach to Detecting Invalid ClicksStrategic ObjectivesPreventive MeasuresCharacterization of Invalid Click DetectionProcess of Detecting and Removing Invalid ClicksPost-filtering Sub-stagesPerformance of Online FiltersIndirect EvidencePerformance of FiltersConflict of InterestAutomated Offline Detection MethodsAlertsAuto-termination System for AdSense PublishersManual Offline Detection MethodsInvestigation of AdvertisersInvestigations of the PublishersResults of InvestigationsThe Lane’s Gifts vs. Google The Lane’s Gifts vs. Google Report part 2Report part 2Andrew HolomanAndrew HolomanGoogle’s Approach to Detecting Invalid ClicksGoogle’s Approach to Detecting Invalid ClicksGoogle’s Click Quality Team’s Mission Statement states:Google’s Click Quality Team’s Mission Statement states:Protect Google’s advertising network and provide excellent Protect Google’s advertising network and provide excellent customer service to clients by: customer service to clients by: –Vigilantly monitoring invalid clicks/impressions and removing Vigilantly monitoring invalid clicks/impressions and removing its sourceits source–Reviewing all client requests and responding in a timely Reviewing all client requests and responding in a timely mannermanner–Developing and improving systems that remove invalid Developing and improving systems that remove invalid clicks/impressions and properly credit clients for invalid clicks/impressions and properly credit clients for invalid traffictraffic–Educating clients and employees on invalid Educating clients and employees on invalid clicks/impressionsclicks/impressionsThe Click Quality Team practices this by raising the levels The Click Quality Team practices this by raising the levels of click fraud detection methods where it becomes hard of click fraud detection methods where it becomes hard and unrewarding to commit click fraud against Googleand unrewarding to commit click fraud against GoogleStrategic ObjectivesStrategic ObjectivesGoogle tries to achieve these objectives in Google tries to achieve these objectives in two ways:two ways:–PreventionPreventionDiscouraging invalid clicking activities on its Discouraging invalid clicking activities on its Network by making life of unethical users more Network by making life of unethical users more difficult and less rewardingdifficult and less rewarding –DetectionDetectionDetecting and removing invalid clicks and the Detecting and removing invalid clicks and the perpetratorsperpetratorsPreventive MeasuresPreventive MeasuresGoogle tries to build mechanisms for preventing Google tries to build mechanisms for preventing invalid clicking that reduce inappropriate activities invalid clicking that reduce inappropriate activities and some of these are: and some of these are: –Making it hard to create duplicate accounts and open Making it hard to create duplicate accounts and open new accounts after the old ones are terminatednew accounts after the old ones are terminated–Making it hard to register using false identitiesMaking it hard to register using false identities–Development of certain mechanisms that automatically Development of certain mechanisms that automatically reduce fraudulent activitiesreduce fraudulent activitiesCharacterization of Invalid Click DetectionCharacterization of Invalid Click DetectionOnline filtering vs. Offline monitoring and analysisOnline filtering vs. Offline monitoring and analysis–Are there some time constraints on how fast the invalid click Are there some time constraints on how fast the invalid click detection should be done?detection should be done?Automated vs. Manual detectionAutomated vs. Manual detection–Were invalid clicks detected by a special-purpose software or Were invalid clicks detected by a special-purpose software or by human expert?by human expert?Proactive vs. Reactive detectionProactive vs. Reactive detection–Has the detection of invalid clicks occurred before or after the Has the detection of invalid clicks occurred before or after the advertisers complaint ?advertisers complaint ?Where were invalid clicks made?Where were invalid clicks made?Were invalid clicks associated with the AdSense or Were invalid clicks associated with the AdSense or AdWords programs ?AdWords programs ?Process of Detecting and Removing Invalid Process of Detecting and Removing Invalid ClicksClicksPre-filtering Pre-filtering - removal of the most obvious invalid - removal of the most obvious invalid clicks, such as “testing” and “meaningless” clicks clicks, such as “testing” and “meaningless” clicks before they are even seen by the filtersbefore they are even seen by the filtersOnline FilteringOnline Filtering – several online filters monitor – several online filters monitor various logs for satisfying these conditions and various logs for satisfying these conditions and detect the clicks in these logs satisfying these detect the clicks in these logs satisfying these conditions; such clicks are marked as “invalid” conditions; such clicks are marked as “invalid” and are subsequently removedand are subsequently removedPost-filteringPost-filtering – offline detection and removal of – offline detection and removal of invalid clicks that managed to pass the online invalid clicks that managed to pass the online filtering stagefiltering stagePost-filtering Sub-stagesPost-filtering Sub-stagesWhat are the two sub-stages of post-filtering?What are the two sub-stages of post-filtering?–Automated monitoringAutomated monitoring–Manual reviewsManual reviewsPerformance of Online FiltersPerformance of Online FiltersA true click is either valid or invalidA true click is either valid or invalidGoogle filters can label a click as valid or Google filters can label a click as valid or invalidinvalidThe actual click vs. click labeling can raise The actual click vs. click labeling can raise confusion within the filtersconfusion within the filters–In finding the accurate number of valid and In finding the accurate number of valid and invalid clicks: invalid clicks: TP+TN+FP+FN=NTP+TN+FP+FN=N–The accuracy rate is equal to (The accuracy rate is equal to (TP+TN)/NTP+TN)/N–The error rate is equal to (The error rate is equal to (FP+FN)/NFP+FN)/NIndirect EvidenceIndirect EvidenceTwo main pieces of evidence for Google’s Two main pieces of