Ad fraud bots, beware — two companies are targeting you with new weapons

Digital advertising is so rampant with fraud, it’s a wonder we don’t see FBI agents raiding programmatic platforms.

In the place of G-Men, two companies — one with a software solution, the other packing hardware — are offering new approaches that they claim go a long way to cleaning up this Wild West.

And it’s a big West. A recent study by the Association of National Advertisers estimates digital ad fraud will cost advertisers in the vicinity of $6 billion this year. As much as 11 percent of digital display ads across all platforms involve fakery, the study said, plus nearly a quarter of video ad impressions.

Today, Santa Monica, California-based Pixalate is facing down the bad guys with hardware.

It is introducing the Security Dome, a single-blade, custom-built, server-like appliance that supplants the company’s previous analytics installed on your system.

Not only does the device and its accompanying Global Fraud Threat Intelligence Cloud service generate a very cool world map with arrows indicating the source and destination of fraud attempts (the top of this page), but it also purports to do the impossible.

It can, the company said, block nearly 100 percent of all kinds of fraudulent actions on mobile web, desktop web, or apps. For ads, those actions drive up fake impressions or ad clicks in order to boost advertiser payments.

VB’s new Brands & Mobile Advertising: How to win report is available for
$499 on VB Insight, or free with your martech subscription

Mobile demand-side platform Airpush, Pixalate VP of product management Khalid Razzaq told me, was able to capture 98 percent of perceived threats using the new device and the service.

The fraudsters have an impressive array of tools for ad and other fraud, which Pixalate says it can swat down before they hit a system. These include suspicious proxy servers, denial-of-service attacks, malware, phishing attacks, compromised IP and URL addresses, and up to 18 different kinds of bots.

The key problem, he said, is that this is too complex for third-party verification services and too fast-moving for software tracking. Ad fraud can get subtle, like hidden ads that are staggered one atop another to show zillions of ads on one page, each in a single pixel.

Customers can buy the hardware appliance, which includes five different fraud intelligence feeds; they can purchase just the feeds; or they can send their traffic to Pixalate for analysis. There is also a free, monthly feed of the worst IP addresses.

Way east in New York City, mobile ad platform AdTheorent is sharpening its software to nab the ad pirates.

Last week, the company announced it had implemented a software-based anti-fraud infrastructure for its platform, after more than a year’s development. The result, it said, is “the first truly clean mobile supply for advertisers.”

Its solution employs machine learning — doesn’t everything these days? — with statistical analysis to spot anomalies that threaten the mobile web and app ads the company provides.

Unavoidable clicks on ads

AdTheorent had been relying on third-party verification services, managing partner and chief legal officer Jim Lawson told me. But, like Pixalate, it discovered they were inadequate.

One reason, Lawson said, is many verification services are more focused on websites than on mobile.

AdTheorent’s ad platform is connected to a number of ad exchanges, he said, but when they flagged problematic traffic and an exchange blocked it, the fraudsters just adapted. The amount of fraudulent ad traffic, he said, was ranging from 10 to 15 percent of impressions.

Some ad fraud, the company said, is too subtle even for its optimized software. So, it has a team of human analysts in its Pittsburgh office giving scores to publishers in the network — and keeping an eye out for such things as pages where a click on the ad is unavoidable.

AdTheorent said that its closest competitors — like Millennial Media, Opera, Rocket Fuel, and Turn — can’t match the amount of data it ingests or the versatility of its analytical model.

“We don’t need to write a new model when a variable is removed,” Lawson said. “The model adjusts.”

Comparing the two companies’ approaches, AdTheorent’s is designed for its own mobile-only traffic, while Pixalate offers a solution available to anyone and for all platforms.