• by Asaf Greiner , May 25, 2018

Ad fraud is costing advertisers far too much money to ignore; $19 billion in 2018, or 9% of total digital advertising spend according to Juniper Research. As a result, everyone is scrambling to find the ultimate mix of processes and tools that will ensure ads are seen by real people.

Blocking bad traffic is a good first line of defense but by itself it’s not a cure. To prevent the fraud from coming back, you need to take a sample of bad traffic and do full diagnostics to identify the source. The broader the approach to fraud detection, and the more methodologies used simultaneously, the higher the likelihood that schemes will be detected. A combination of blocking and performing traffic diagnostics is the best way to ensure that fake clicks and views won’t drain advertisers’ budgets both today and tomorrow.

Detection systems can be duped

Blocking bad traffic works, and it is essential for helping ad measurements better reflect reality. But blocking fraudulent traffic is like antibiotics — in the beginning the treatment is effective but in time fraudsters can adapt and become stronger, making the treatment ineffective.

Based on the settings that trigger alarms for anti-fraud systems, fraudsters can change their measurements to make their fake traffic appear real the next time around.

Making fake traffic appear legitimate is quite easy to do. IP addresses can be changed when a large number of clicks are coming from a data center, or any one single location. The URLs of long tail sites can be changed to appear as premium tier one news sites. Locations can be faked to hide the fact that an abnormally large number of clicks come from the same geographic area. CPM and CPC, are particularly susceptible and easy mimicked by malicious bots and fraudsters, for example, fraudsters can keep CTR under 0.3% to appear in line with expected KPIs.

Methods used by many fraud prevention systems have become so predictable that black-hat publishers can purchase traffic specifically designed to evade the fraud-detection algorithms of popular prevention vendors.

Working towards a cure

Blocking is effective, and it might be the only weapon to use when protecting a premium advertiser with a zero tolerance policy for ad fraud but is just one weapon to utilize in the battle against ad fraud. If the goal is changed from just cutting out bad traffic to finding the source, there is a higher likelihood of a long-term cure.  By letting in a sample of bad traffic and then performing an analysis to find out who is responsible for the fraud marketers can have a better understanding of the nature of the problem and how to solve it. Fraud experts leveraging machine learning techniques can find the source and work with partners to clean up traffic for the long-term. 

Periodically sampling and performing diagnostics is also a good way to cross check ad blocking systems. Without confirming that there is real fraud false alarms could be triggered that force marketers to cut off publishers unnecessarily, damaging long term mutually beneficial business relationships.

Combining reactive and preventive measures also eliminates the need for too draconian goals for clean traffic, for example zero fraud, which can result in limiting traffic sources to only the most selective and expensive publishers pushing down digital advertising ROI. Traffic blocking combined with analysis keeps fraud levels down while making digital advertising more cost effective by pinpointing with accuracy the portion of traffic that’s fake and eliminating it. 

The adoption of both ad fraud prevention and detection will keep fraudsters on their toes, making them work infinitely harder to sneak in bad traffic.

Ultimately, the long term goal is always to strengthen an immune system to fight a disease, instead of relying on antibiotics for the long term. Antibiotics aren’t always the correct treatment and even when they are used effectively, over time bacteria develops an immunity to the treatment. Similarly, by exclusively blocking bad traffic without diagnostics, the whole system is more vulnerable, and can let the fraud disease rage on and on, with fraudsters always staying one step ahead of the anti-fraud solutions.