Yesterday we posted a story about fraud detection tools that has affectively stirred the performance marketing pot. Originally, CPAWAY’s Thomas Dietzel made a fiery statement claiming that these tools have been costing him revenue by scrubbing legitimate leads—an obvious issue for any performance marketer. We also heard opinions from Brian McLevis of Scrubkit and David Sendroff of CPA Detective, both of whom added their perspective to the mix and giving us all sides of the story.
However, the debate doesn’t end there, and this post will attempt to bring you arguments from all across the board.
One of the earliest comments that sparked further discussion came from Matt of Smarter Choas, who supported the use of these tools, but with careful, human supervision. You can read Matt’s full response here.
The main point of his argument:
“In the end, work with those companies that have enough expertise and brains to use such a powerful tool, and collaborate with those companies that are trying to protect our domain to make this a better space.”
In response to this, we heard once again from CPAWAY’s Thomas Dietzel who offered up an in-depth rebuttal:
My issue isn’t the tool or that tools exists. We have similar tools in place. My issue is any advertiser and/or network that uses tools as a sole reason for not paying as you noted with your North Korea analogy. We just want the tools to educate their customers to use human common sense when reading reports.
Since my post, it seems that some tools allow human common sense to be used (being transparent with data points and reasons) and others give a blind thumbs up and thumbs down. Human intervention is the key and replying on a program doesn’t always work, no mater what they claim.
Even with incentive traffic there is good and bad. When we receive a report saying anything is bad, we look at the flow of the offer and data points we have captured (the same that all other programs capture, if not more as we service the click). If the flow of the offer didn’t back out, that is not something we can control when you accept this type of traffic. If it didn’t back out due to true fraud, then we will review it and ask for documentation. The PRIMARY point of my post and response here is: “Don’t tell us Because “INSERT SOFTWARE NAME” told us so” as the reason. We need to know specifics. This goes back to your analogy, too many people are using these tools as a reason not to pay, when they are getting paid and/or have been able to use the data with no problem, and furthermore their merchants haven’t even complained to them.
A prime example of it is, and one that triggered my outcry on Facebook is someone we worked with for 6 months, never an issue, always asking for more, no traffic pattern has changed and/or sources. They implemented CPADectective, and then sends a huge reversal file to us stating, they are resulting in a high risk score and they don’t want to pay as a result of such.
When we asked for data and reviewed all the points on our side, it looks fine to us and that their coreg path on the back end simply didn’t back out for them this month… which as a result they lost money instead of profiting….
We review every IP that comes in for proxies in real time on the click and so forth. We can’t stop friendly fraud, but we can stop the bulk of the hard core true fraud in real time. We compare patterns of traffic, a couple examples include:
1) if 100% of the traffic is coming from one ISP it is a concern.
2) if 50% of the leads are coming from a Linux Operating System, its a concern when normal is always under a couple percent.
3) Browser Language on all leads isn’t EN when its a US offer
Concerns like that the above are TRUE fraud and not friendly fraud. We are capturing EVERY possible data point on every click that comes in for this simple reason, so we can compare the data.
In regards to PR approach one company took:
If you claim your product is perfect, you are absolutely fooling yourself and your customers. We like everyone else, are very good at catching the bad apples… do we catch them all no. Do we have false positives, that require human intervention to read the reports? Yes!
My point to everything is: USE COMMON SENSE, REVIEW THE DATA. If you do not know what the data means, perhaps reach out to someone in this space you can trust that KNOWS what the data points are and what to look for. Don’t rely on a blind thumbs up or a blind thumbs down. If you don’t have the ability to ask, perhaps impliment both ScrubKit and CPADectetive.
Just note, these tools are only as good as the data points that can be passed to them by your tracking platform. Not every tracking platform can pass them the same data points, so each tool has its pro and cons. When in question and within your budget use both, compare, then if they don’t provide same answer on if something is phishy or not, question them both why…
Networks and Merchants need to be able to back themselves up with FACTS when they cry fraud. Simple as that. If you can’t provide facts that are indisputable, then its just another tool for not paying out and keeping 100% margin.”
Surely this is not the end of this discussion, and we look forward to hearing more views on the subject and digging into the deeper issues at hand. Whether you’re a publisher, a network, or an advertiser, TheMail wants to hear your opinion on these tools—let us hear your comments below.