But even with the higher visibility and greater scrutiny of online ad fraud, it seems to be a problem that only gets bigger.
The most recent example of the phenomenon came to light a few weeks ago, when ad fraud prevention consulting firm Pixalate announced that a newly discovered botnet has been draining literally billions of dollars from advertisers’ MarComm coffers.
Xindi is making money for its creators by serving actual ads – but to simulated audiences. It has spread via familiar methods such as phishing.
Pixalate estimates that just shy of 78 billion fake ad impressions have been racked up so far. Even at low cost-per-impression revenue figures, the high volume amounts to several billions of dollars of illicit revenues siphoned (and counting).
What makes the Xindi botnet particularly nettlesome is that it’s designed to go after computers and networks at high-end organizations, enabling it to “mimic” desirable web traffic (i.e. affluent consumers).
According to Pixalate, already there could be as many as 8 million computers compromised in more than 5,000 networks, including a goodly number of Fortune 500 companies as well as university and governmental networks.
Such desirable locations and ad audiences translate into lucrative online ad pricing (CPMs of $200 or more).
In the event, advertisers are paying high prices … for nothing.
To counteract Xindi, Pixalate recommends that the Internet Advertising Bureau update its protocols to factor in the pace of ad requests, so that impression generated after a certain time period cannot be accepted as valid — and hence would be non-billable.
Whether this or other remedies will actually happen is up in the air at the moment (the IAB isn’t onboard with the recommendations).
Either way, what seems clear is that whatever the remedial actions that are taken, burgeoning ad fraud activity is bound to continue.
The question is, can it ever be contained, or will it just continue to grow and grow? If you have any thoughts or ideas on the challenge, please share them with other readers.