Online ad fraud and ad viewability are two of the hottest topics in advertising these days. Chances are, you’ve seen the alarming headlines: Up to half of online ads are never seen by humans, they scream; Over $6 billion could be lost this year to ad fraud.
What does this mean to you as an advertiser (or simply as a consumer and internet user)? First, it’s really happening and can’t be ignored. Second, the sky isn’t falling; there’s no need to panic just yet.
Viewability and fraud are actually two separate issues that have been conflated together, mostly because they make better headlines that way, with higher and more shocking estimates of the financial impact. And some of the organizations whose PR machines are pushing hardest for those shrill headlines are the companies now selling technologies that promise to “automatically detect and eliminate” fraudulent ad views or clicks.
This is a bit like the fox guarding the henhouse, because it’s the rise of automated technology in the ad delivery process that has been primarily responsible for the problem these organizations are offering to solve.
How Ad Fraud Works
In the early days of internet banner ads, agencies would scout out prime websites on which to place their clients’ ad messages, and make a deal with the publisher to pay for exposure to their high-quality audience on a cost-per-thousand impressions (cpm) basis. This still happens, of course; and it’s a great way to deliver ads to a premium audience through a trusted publisher.
But the wonders of the internet have given us new ways to reach targeted prospects. Thanks to cookies and other tracking mechanisms, every computer with a connection to the world-wide web now has an online profile that says where it’s been and what it’s done. In the jargon of marketing mavens, these are assumed to be profiles of consumers—but in fact, they’re profiles of computers, which are assumed to be controlled by individual users.
So here’s a quick explanation of the most prevalent—and troublesome—type of online ad fraud.
1. The fraudsters start by using a virus to gain control of thousands of computers.
2. Next, they instruct those computers to visit thousands of websites, racking up page views and clicking on ads. But at this point, all they are doing is creating profiles of fake consumers that will appear highly desirable to various marketers. For instance, a computer that has visited dozens of cycling stores and bicycle club sites represents a prospect that would make any high-end bicycle manufacturer drool. But in this case it’s not a real person, it’s a bot.
3. Now comes the really clever part. The crooks create a website with an actual URL and a few pages littered with bike-related keywords, but no readable content a real person would want to visit. Because they can drive their own bot-traffic to the site, it looks good to the exchanges, so they are able to get it listed and offer their fake bot traffic as “inventory.”
4. Now they’re in the game, collecting money for ad views on their site by computers that have great profiles, but no humans behind them.
This is a double-whammy for the online ad industry, because in addition to the false views and clicks the crooks are being paid for, the scheme starts by sending bots to many high-profile websites run by reputable publishers, and in the process inflates their visitor numbers dramatically. Which means advertisers could be paying for false clicks on those sites (although the fraudsters themselves may not be profiting from those impressions), and the overall traffic numbers become suspect.
Blink and You’ll Miss It
Clearly, this is a real problem for the industry. Although fraud has been around since approximately the day the first ad appeared on the internet, it has become a huge issue in recent years due to the growth of Real Time Bidding (RTB).
RTB is a complex process that auctions off page views to the highest bidder, and it takes place in a time frame that makes the blink of an eye seem slow. A user/computer clicks on a link and a web page starts to load. The “profile” of that computer is now up for bids. Based on its cookie history, it will fit the target profiles sought by various advertisers. The highest bidder’s ad is then served to the page in real time, as it loads.
According to Ben Kneen of Ad Ops Insider, this whole process takes about 50 milliseconds (that’s 0.05 seconds). It is replicated around the world millions of times an hour, and obviously can only be controlled by an incredibly sophisticated array of computers. Hence, the need for still more computerized technology to monitor the process in attempting to weed out the fraud.
Now, back to that question of what this means to you, and what you can do about it.
Don’t lose sight of the fact, first of all, that RTB is a real opportunity for advertisers who want to connect with well-targeted prospects on the internet. The extent of the fraud is variously estimated at below 5%, or north of 20%, but that depends a lot on how you set up your campaigns, and how selective you are prepared to be. If you’re Kraft Foods, you can apparently be very selective. Julie Fleischer, Kraft’s director of data, content and media, told Ad Age recently, “We’re rejecting 75% to 85% of the impressions available.” However, if you don’t have the vast resources of Kraft Foods, spending $35.9 million a year on digital advertising, you may need an agency champion on your side, staffed with people who know the ropes.
Also, don’t trust your whole budget to the automated bots. RTB can be a valuable part of the mix, but should never be all of it. Any agency you work with should provide a balanced plan, with a rationale that ties it to your business goals. Ensure they monitor the progress of each campaign, and know how to read and interpret the analytics.
In other words, if you don’t want to be robbed blind, don’t wear a blindfold!
Next: A Look At Viewability
Viewability is another issue, which Google has been taking a leadership role in both defining and tackling. I’ll discuss the good news side of this challenging problem in my next post.