• Abhinav Bangia

What was invalid yesterday is valid today, the labyrinth of Ad Tech.



Data isn't just the source for anomalies detection for advertisers today when it comes to fraud, but it is also the source of reverse engineering for fraudsters to bend the rules and mimic bot traffic as real traffic.


If you have been sharing ad fraud data with your partners, you are fueling a much bigger problem than it was at the first place. Every time a type of invalid traffic reason tags are shared within partner network, the clown in the same partner network works to find the way around so that the invalid traffic today becomes valid tomorrow.


Basically, invalid traffic is any activity that doesn't come from real user with genuine interest. We are working round the clock to break this systematic type of fraud chain and make sure people who cause it don't profit from it. We are working towards an compliance with Interactive Advertising Bureau , Media Rating Council, Trustworthy Accountability Group to build world's first bait-caught model, where certain traffic changes are create deliberately to understand who is plaguing the supply chain.


If you are committing to any of the 4 mistakes below, you need to probably rethink again.


1. Real publishers/networks never ask for attribution data for optimization.


If you are being asked for post acquisition/attribution data from your trusted partner network in the name of optimization of your campaigns, you are probably being fooled around. The fraudster would eventually look into how does real traffic for your application looks like in terms of attributed data and mimic all it's sources to fudge all your matrices. This would be the death of performance which would impact your business certainly in 2-3 business quarters.


2. Never let an acquisition partner run your re-targeting campaigns.


If you are already doing this, you might have to rethink this particular strategy. Your re-targeting audience is actually your loyal customers that is your brand's real food, you would never want a competitor's promotional creative to reach their eyes which can actually slip this particular customer into your competitors pocket. You might risk your CRM to be used for your competitors acquisition campaign. Set a neat line of legal understanding with your partners on usage of this data only for re targeting and not fuel any acquisition DMP whatsoever.


3. Keep changing your fraud detection algorithms.


The fraudster's intention is to understand what logical filters or algorithms you are deploying to understand invalid traffic in your campaigns. Once this understanding is developed, it takes seconds to deploy invalid traffic mimic real user traffic. You would need to collate bunch of algorithms at your place and keep using them in mix-match fashion.


4. Not storing partners past mischief's.


Every organization and network in the world today is plagued with ad fraud. It's alright to understand your partner's stand to improve over time, but it is foolish to not store the invalid traffic that you had caught before under partner's name/sub ID name table. Over the time, this data would help you understand the systematic name change of Sub-IDs but same fraud behavior in the subsets.


Without facts and principles, data is useless.” – Bob Hoffman



57 views

Com Olho is a leading cloud-based machine learning as a service platform. We make it easy for CXO's to understand their consumers using behaviour analytics, machine learning and data science. 

e : contact@comolho.com

© 2019-2020 COM OLHO IT PRIVATE LIMITED

  • Facebook
  • Grey LinkedIn Icon

Made with      in India