Features

Programmatic Advertising’s Dirty Little Secret Gets Bigger: Fraud is on the Rise

Written on
Dec 30, 2013 
Author
Andrew Pancer  |

EDITOR’S NOTE: This article, originally published on April 18, 2013, placed at No. 6 in our 20 most popular articles of the year.

ADOTAS – Programmatic buying should be a boon for advertisers. It should be a powerful tool for them to reach relevant audiences at scale in an increasingly fragmented media environment. In theory, it increases the accuracy of their targeted advertising efforts while eliminating wasted spend. There is a plethora of quality inventory from legitimate publishers available through programmatic channels such as SSPs and ad exchanges.

Unfortunately, the enthusiastic adoption of programmatic buying by marketers has brought with it some challenges. Non-Human Traffic or NHT is a growing problem that is polluting the online advertising ecosystem. Integral Ad Science (formerly AdSafe) classifies roughly 30 percent of all exchange traffic as suspicious, exhibiting the traits of NHT. Brian Pugh, SVP of Audience at comScore, estimates that NHT has grown six times since 2011.



While these numbers are alarming, let’s put the problem in perspective. In the same way that spam plagued the nascent email marketing industry and click fraud ran rampant in the early years of search engine marketing, NHT is a clear and present threat to programmatic advertising. Issues with spam and click fraud were largely resolved once marketers and responsible vendors turned their attention to rooting them out. But not before a lot of marketers lost a lot of sleep — and a substantial portion of their advertising budgets. We can only hope that the same will be said for programmatic, and soon. How quickly we can all look back at the issue remains a question.

Let’s examine how the fraudulent activity occurs. NHT originates with a network of computers that have been compromised by malware. Each of these infected computers is known as a “bot”. The distributor of the malware (a “botnet”) can access the infected bots and control their activities, most of the time without the knowledge of the innocent owner of the computer. The botnet is paid by website owners to manipulate the infected browsers and drive traffic to those websites, all invisibly and without the knowledge of those users. These website owners, or “ghost publishers,” are able to generate billions of advertising opportunities in a very short period of time. Typically, the ad impressions produced by these bots are offered for bid in advertising exchanges, and generate illicit revenues for the ghost publishers.

All of this fraudulent activity produces opportunities for revenue-generation by otherwise legitimate players up and down the food chain in display advertising. Ad exchanges, DSPs, supply-side platforms, media agencies and others all can benefit from the funding of these specious impressions. Whether their involvement is unwitting or just complacent, the increased media spending and fees generated are largely responsible for the mostly toothless industry response to fraud to date. For some media buyers, as long as they’re meeting their key performance indicators (KPIs), they have little incentive to investigate whether impressions are delivered to actual consumers. Similarly, many intermediaries abdicate any responsibility for verifying impressions beyond the reporting they receive from the channel.

Marketers, of course, are the ones who suffer. If their campaigns are served to websites that load in the background and are never seen, then their budgets are wasted. Recent studies suggest that this kind of fraud wastes tens of millions of ad dollars per year.

At m6d, we have seen a massive rise in suspicious activity over the last 18 months, and that suspicious traffic comprises a significant percentage of what our bidder sees every day. We have developed technology to detect suspicious sites, and eliminate them from our bidding universe. We have found that the sell-side is not yet equipped to do this filtering themselves, and we have started to train them to eliminate fraud before it gets to us.



Co-visitation networks identify sites that have large amounts of non-intentional traffic. These clusters represent sites that have over 50% overlap traffic (i.e.: YouTube, NYTimes since they cater to a mass audience). Clusters of bright red represent sites with ABNORMALLY MORE than 50%+ shared traffic.  Some of these sites are being built just to shepherd traffic around and monetize. Over the past 2 years, there has been an explosion in this this type of fraudulent activity.

There is some debate as to whether the buy side or the sell side should take the lead in weeding out suspicious inventory. We think both need to take action. The buy side needs to come up with better KPIs to root out the opportunity for bad actors. An increased emphasis on viewability as a KPI could help shine a light on NHT; after all, an ad that is never seen cannot hope to influence purchase decisions, which should be the ultimate measure of success. A wider adoption of multi-touch attribution would be a huge positive step for the industry. Rather than simply attributing credit to the last impression to touch the consumer before a purchase, allocating credit to multiple touch points along the consumer decision cycle provides a more holistic view of marketing performance. And as a by-product, it will expose fraudulent practices.

While the buy side can take steps to identify NHT and can demand higher quality for their client’s campaigns, it is the sell side — the exchanges and the supply-side platforms (SSPs) — that truly have the power to combat fraud. Fortunately, it is in their best long-term interest to do so. It is crucial that they improve their certification protocols to determine the legitimacy and quality of the inventory placed into their exchanges.

With spam and click fraud, eventually tools and systems were put in place to identify and minimize the threats, and now we don’t hear too much about them. We are hopeful that this story will have a similarly happy ending. Both major players and enterprising startups will create the tools necessary to foil fraudulent tactics. Until that happens, it is urgent that the industry come together to demand transparency and validation of traffic. Those who either practice or condone fraudulent behavior will be on the wrong side of history, and their reputations — and businesses –  will suffer for it.





Andrew joined m6d as COO in September 2008. He was Vice President of Digital Development at the New York Times, and COO and CFO of About.com, where revenues exceeded $100M and profits tripled during his tenure. He has a BS in Business Administration from Washington University and an MBA from NYU.

Reader Comments.

Great article Andrew! Beyond the fact that RTBs and Exchanges are, IMO, driving eCPMs down, when you load on fraud – where are we heading?

Posted by Lance Jackson | 1:53 pm on April 18, 2013.

This article is a bit confused.
Bot nets are human. The traffic originates from a user’s computer. The users infected look at (i.e. viewable) and interact with (i.e. mouse over’s and clicks).

Bot nets still work in search. Search Zeus for a bot net recently discovered by Microsoft.
The great bots mimic human traffic perfectly.
Fake traffic shared between fake sites is a different problem
The real lesson…you get what you pay for. If you buy RTB … this is one of the costs

Posted by john simpson | 2:48 pm on April 18, 2013.

Ironically, I just touched on many of the topics discussed in this article in my blog post yesterday at http://danielbackhaus.com/advertising-that-works-sorta

Both click fraud and phantom traffic remain current problems and their occurrence is widespread and rampant. They afflict virtually all segments of digital advertising (with exception of directly managed premium) and are by no means limited to programmatic campaigns. In the lower-priced, remnant categories they do occur more frequently, but there is no reason – other than apathy, careless media buyers and campaign managers, and misaligned incentives for this state.

Posted by Daniel Backhaus | 8:49 pm on April 30, 2013.

While fraud is certainly an issue, I always find this discussion interesting by comparison to other media outlets.

Neilson & Arbitron ratings certainly don’t take into account viewers and listeners that are out of the room when an ad plays (or zip through them with their DVR). Magazines and Newspapers don’t take into account the “subscribers” that only read the publication for one topic (e.g. sports) and miss the ad that was placed somewhere else.

The disadvantageous advantage of our media is its ability to track activity in minute detail. Perhaps this is one reason many (most?) marketers really haven’t put up any fuss as long as they are meeting their KPIs. To that point, even in the cleanest of environments, if you aren’t meeting your KPIs you drop that program.

It all boils down to one simple word — value.

Posted by Major Webuser | 6:04 pm on December 30, 2013.

Leave a Comment

Add a comment

Tags: , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and
Article Sponsor

More Features