I’ve been tracking down fraudsters for quite a bit, first for the US Secret Service, and then in the advertising industry and also in my publications. Nothing really irks me so much as fraud --- especially when it’s played by some of the largest industry players in the land.

The fraud not only makes our entire industry look worse, but makes it harder for legit, honorable companies to break-through when the "Titans" of industry are taking advantage of everyone.

This problem has been exacerbated in recent years by the rise of programmatic advertising which replaced the old network-publisher relationship business with a more automated approach where machine-based ad exchanges control the ad placement process without human oversight.

This led to a massive drop in price and quality.

Now an estimated 70% of online adverts are never seen by humans.

This is hard for me to believe, but even harder for me to excuse.

To show how useless programmatic display can be when resold over and over again, Dr. Augustine Fou bought 1 billion ad impressions, and despite a million clicks, didn’t get a single conversion. I want you to think about this – supposedly about one million people clicked on his test ads, a billion people saw the ads, but not a single person converted.

As someone who has bought trillions of impressions in the last two years, I can tell you that this is absurd as it sounds.

According to Dr Fou: “As more ad inventory is bought and sold programmatically on ad exchanges, bad guys are finding it far easier to commit fraud because few agencies and advertisers actually check in detail the hundreds of thousands of sites on which the ads are run. It’s easier to hide in a far larger haystack.”

If this makes no sense to you, join the club of experts, because it’s obvious fraud, but the network’s “fraud detection” didn’t catch anything, even though a third grader could tell you something was extremely fishy.

“Fraud goes to zero when every deal is sold directly, whether it’s through an IO or [guaranteed] programmatic,” said Matt Prohaska, CEO and principal of Prohaska Consulting  in Digiday. “It’s the SSPs who are doing an awful job sheltering and allowing criminals to just set up shop.”

Yes, some of the SSPs are helping the fraudsters commit fraud. 

So you’d think the advertisers and brands had a solution, right? Nope! In a recent survey of advertising decision makers, 43% said that they could not estimate how much of the suspicious traffic on their websites was originating from sophisticated bots, and had no solution internally to even stop it.

So why is this happening? Why is there so much fraud? I have more than a few ideas why, but here are some of my most prevalent theories.

1) Everyone is looking for a deal, and this promotes fraud. Still to this day, many agencies look to lower the effective CPM or provide more eyeballs to make it seem like they are providing their clients more inventory and father reach. However, often these “deals” are nothing more than junk inventory that no one will see, let alone convert. There is no doubt in my mind many of the agencies either look the other way or just plain ignorant.

2) Many Fraud Detection Companies are USELESS. Minus a few exceptions, most of these companies that claim they can prevent fraud through a complex algorithm that sniffs out the worst offenders and prevents them from defrauding your company but do little or nothing. The sad fact is that most of these companies are not used to prevent fraud, but to make advertisers feel better about their buying decisions. Worse, almost none of the fraud systems have a straight definition of fraud and allow their clients to “move the fraud slider” to allow as much fraud or little fraud they want.

3) Programmatic Networks Don’t Want to Solve Fraud. Here’s a hard thought to stomach: some of the largest ad networks really don’t care about fraud, and only do enough to make it seem like a “hearty effort.” Everyone is looking for billions and billions of advertising impressions now, and there is no way you can buy that much without significant fraud. Now the networks like Gannett, to fill these orders, are allowing ads to be shown anywhere. There is no way that any new “large” networks are anything but reselling junk.

4. All these Vanity Metrics are Junk Science. In  order to keep the money coming, more and more networks are creating new metrics that have very little real-world value. Many of these strategies are based on pseudo-science and little regards to data-driven testing, measuring, and learning – it’s a poorly crafted pastiche of buzzwords from hasty collected client inputs.  It’s almost all meaningless dribble. Any campaign should have some metrics that can prove that people are real, such as popping a survey up once in a while.

I think we need to make this clear: this has gone from “occasional mistakes” made by some large networks, to what seems to be a clear cut case of some networks and technology partners working together to defraud advertisers to the tune of billions of dollars.

This isn't where our industry should be going, ever.

 If this continues, expect the FTC and perhaps event federal law enforcement to start investigating some ad firms as nothing more than crime syndicates. 

Online advertising has become the gold standard for companies around the world — sometimes replacing the traditional TV, radio, and print ads. Part of the reason for this is the sheer cost-efficiency of online ad campaigns compared to traditional ads on television.

With TV ads, companies have to pay for air time on different networks, set up a whole production with paid actors, scriptwriters, directors, makeup and special effects personnel—the whole nine yards. 

Online ads, especially banner ads or sidebar ads on other websites, are much less costly by comparison. In fact, the average cost per lead for traditional advertising channels is about $619 while the cost per lead for programmatic display ads is about $38 (Source: Linchpin SEO).

So, if programmatic ads are so much more cost-effective, what’s the problem? Who wouldn’t want to be able to get roughly 16 times as many leads for the same ad spend?

The issue is that there’s a dirty little secret about programmatic display ads that the platform providers selling those ads don’t want you to know:

The prevalence of ad fraud in the industry and how little gets done to stop it.

What Are Programmatic Ads?

Normally, when advertising on someone else’s website, you would request a proposal, negotiate with a person, and trade quotes and modifications back and forth with a real person. Programmatic ads automate this process.

Instead of negotiating with a person, a programmatic ad system uses software to determine when and where to place ads in front of a specific audience. These software algorithms use a variety of rules and attributes to determine the best ad placement. It’s a big time-saver for people who have more job responsibilities beyond ensuring that every ad placement is perfect for a single ad campaign.

People looking to buy ad space may use a demand-side platform (DSP). Meanwhile, organizations looking to sell ad space may use a supply-side platform (SSP).

About Demand Side Platforms

DSPs are software platforms used by advertisers to buy advertising space on other websites. Many DSP users employ impression-based revenue models (meaning they pay out based on how many people “see” the ad).

This is one reason why many DSP providers leverage viewability as a metric when checking for ad fraud—going “hey, look, it’s viewable, so obviously it’s legit.”

However, this isn’t always the case.

About Supply Side Platforms

SSPs aggregate high-end traffic sources to monetize them through the DSPs used by programmatic ad buyers. Their goal is to sell as much ad space as possible for as high a cost as possible. This doesn’t necessarily motivate them to ensure that the ad space they’re selling is from legitimate sources.

Fraudsters have been known to use domain spoofing to disguise their fake, low-quality website as more valuable websites to trick SSPs and their customers. The advertiser ends up paying way more than they should for ad space on a fake website that will probably either hurt their reputation or only generate fake leads (or both).

Examples of Programmatic Advertising

Programmatic ads come in many forms. Two of the most commonly used types of programmatic ads are display ads and video ads.

Display ads often take the form of a banner or a box on a website. You may have seen examples of this when browsing different websites — like a banner advertising computer parts on a magazine website.

Video ads often get put on social media sites that support video content — such as YouTube or Facebook. Video ads that play at the start of another video are sometimes referred to as in-stream ads, while videos put at the top of an “up next” feed are called in-display ads.

In both cases, programmatic ads are presented to a viewer based on some kind of parameter related to that viewer. For example, a computer parts ad might appear to people who have shopped for computer parts recently, visited a “how-to” site for building a computer, or subscribed to a computer magazine or similar website or blog.

In theory, this enhances the effectiveness of the ad by targeting someone who has demonstrated an interest in the product or service they’re being offered.

The Dirty Secret about Programmatic Ad Fraud Your DSP Wants to Hide

While programmatic ads can be highly effective and cost-efficient, it isn’t all sunshine and rainbows in the programmatic advertising marketplace. A lot of DSP and SSP providers pay lip service to stopping ad fraud and only presenting you with valuable lead sources.

Some even boast that they’ve added pre-bid fraud detection solutions to stop ad fraud in its tracks. The problem is that the so-called “pre-bid fraud solution” is only geared to detect GIVT, not SIVT.

GIVT, or “General Invalid Traffic,” is a term for simplistic bots that are easy to detect (often because they’re meant for innocuous tasks like crawling websites so search engines can index them). Meanwhile, SIVT, or “Sophisticated Invalid Traffic,” is the true threat since it’s almost exclusively used for fraudulent purposes.

As a result, an “ad fraud solution” that only filters out GIVT can’t be considered a full solution to ad fraud. Case in point: Some of my customers in the past have discovered that, for the leads that they got through their “safe” programmatic advertising networks, they still had 35% to 40%, or higher rates of ad fraud.

In fact, one client had an ad fraud rate of nearly 90% — fraud that the pre-bid solution missed, but our ad fraud solution caught.

Why Viewability SUCKS as an Ad Fraud Prevention Metric

So why are so many pre-bid fraud solutions only able to stop GIVT and not the SIVT that fraudsters actually employ? 

One of the problems is that they rely on only one or two data points to determine fraud, which is not enough to identify sophisticated attacks.

Take, for example, viewability.

On paper, it sounds effective. DSPs and SSPs want ads to be viewable.  However, the definition of “viewability” is that a minimum of 50% of the ad was "in view" for a minimum of 1 second. So, people presume that it must have been seen by a real person! 

However, viewability alone doesn’t tell you who (or rather, what) “saw” the ad.

Technically speaking, a display ad that’s one pixel tall and one pixel wide could be counted as “viewable,” even though it might as well be invisible to the naked eye. Plus, nobody’s ever going to click on it outside of a possible accident. This is a common ad fraud tactic known as pixel stuffing.

To make a long story short, viewability is susceptible to fraud because it:

1. Doesn’t have a way to detect fraud
2. Doesn’t provide actionable data to improve your marketing
3. Doesn’t offer any context for who or what is doing the “viewing”