When CPA networks are finding margins tough, it’s easy to try and rely on software tools that do the work for you. CPA Detective and Scrubkit both have their proponents and detractors, and I have spoken to both of their owners at length about their products. However, blindly implementing them and relying entirely on their default presets can cause issues. (photo by Steve Hall)Thomas Michael Dietzel, CEO and Founder of CPAWAY, recently pointed out on Facebook—in a sponsored post—that he was having leads, delivered by incentive publishers, being accepted because the advertiser was using CPA Detective which was automatically counting incentivized leads as fraud.
When asked about this, Dietzel stated,
“Any network or agency that uses any fraud prevention system to withhold funds from a publisher, that in return they were paid for, should be simply ashamed of themselves.
They should also be ashamed of themselves if they allow their merchants to do the same. If the leads are bad, by all means reverse them, but provide documentation. However, simply saying ‘CPADetective says its bad’ is not a reason for non-payment if you were able to work the lead. This has become a huge factor in this space. It has also become an expense for us, with us having to enable Maxmind on every click that goes through our network with certain agencies, and network at 0.004 on average per click to prevent them from crying fraud due to the several platforms out there that use the same data source as a sole identifier.
The industry has changed positively as a result of awareness, but without education on how to use these tools or read the data, it creates an issue for innocent bystanders. For example any anti-fraud system that gives you thumbs up or thumbs down without a transparent reason is not one you should NOT be using. You need something that is transparent which explains the reason for such speculation. The only one on the market that is transparent—that I’m aware of—is Scrub Kit. They however, much like others, for some, not all, of their data rely on Maxmind though like others out there.”
As Brian McLevis of Scrubkit noted,
“You are correct that incent should have it’s own set of algorithms. This is one of the reasons we stepped away from a ‘set in stone’ default rule set. We allow the user to change their settings to adjust for incent offers and assign settings to incent offers. This allows networks to rid false alarms for incent traffic.”
David Sendroff of CPA Detective added to the discussion by explaining,
“We do not flag an entire source because there’s fraud somewhere in the source. We score each conversion based on a number of criteria and then show an aggregate of the percentage of the total that had a ‘high-risk’ score. We provide plenty of transparency to our clients so they can evaluate the exports to understand what’s happening. It’s up to the clients to use their discretion when deciding what to show their pubs for the chargeback. Also, you should know that we have a 1% false positive rate, which is closely monitored by our data scientists to maintain accuracy.”
Traffic quality is an issue and a continuing concern to all—networks, advertisers, and publishers.
However, it’s not all bad news for publishers. Last week while I was at the Diablo Media offices in Denver, their team explained how they were able to help publishers because CPA Detective helped them keep traffic to advertisers clean, increase payouts, and eliminate bad publishers. I guess it’s all about using the tools—not being one.
NOTE David Sendroff says he will provide a full response tomorrow.
We would love to hear your comments below.