When Bots Attack: Media Geeks Combat Fraud

bots

The promise of digital media is in its ability to easily measure and optimize our marketing programs. Without the appropriate measurement infrastructure in place, however, marketers oftentimes fall victim to the illusion of strong performance. What often happens is, the numbers being surfaced are averages and aren’t representative of what most customers are experiencing. Hidden inside are the un-surfaced truth and inefficiencies. As an industry, our misrepresented media metrics that have us feeling good about our investments are actually leading to negligent spend that has opened up the doors to fraud.

Now, financial fraud is older than the 2000 year age we live in today; prominently gaining notoriety in Greece 300 B.C by merchants looking to insure their ships at sea, selling the cargo, sinking the empty ships and getting a second pay day. Most recently, in the young field of Digital Media, we were hit by “Methbot.” A name coined by White Ops, a cyber security organization focusing on fraud and verification that uncovered it. It would be a feat to remove fraud from our industrial underworlds though we can combat it and continue to reduce its impact.

What Happened:

During its run, Methbot stole $3 to $5 million in media revenue daily. Much of the press was focused on the impact it had on video inventory, but Methbot also impacted display. Particularly in early November 2016, which was seen in Just Media campaigns across 6 different programmatic partners reaching very distinct niche audiences. For one partner, Viewability dropped 20% in one day.

This decline continued down a total of 40% with more than 80% of total impression no longer viewable. We worked diligently to improve performance every day, but we didn’t see positive change until a week later. Then, the following day another partner took a hit, followed up again with another partner dropping in the same fashion. One partner launched mid-campaign and was never able to get above 30% Viewability for any given day and had to be removed from the mix.

What We Did:

Our intuitive optimizations yielded lower performance. At the time we did not know Methbot was driving the tumultuous inventory and media delivery issues. Nor did we know how it was targeting trusted and premium inventory. One of the first moves we made with each partner was optimizing the inventory we ran on from the top 25% most viewable to the top 10% most viewable. This means the only inventory we’d bid on and therefore run our campaigns on was tracked as having the most viewable inventory over the remaining 90% of the inventory in the exchanges. Ranked, top tier inventory didn’t yield top results against Methbot.

Where We Are:

Since its revelation, White Ops has publicly shared spoofed domains, a full list of URLs, compromised IP addresses and IP ranges to support the industries response against Methbot. Established partners will work in tandem with Cyber Security partners like White Ops, Moat, IAS and Double Verify to leverage signals on a pre bid basis improving the quality of the media we buy. Viewability for our campaigns have returned to industry averages and above. Regardless, if the industry wants to combat fraud we’ll need to do more than tweak our campaigns and make marginal improvements.

A Call To Action:

As an industry we need to redefine how we’re buying and measuring media as well as how we speak to performance. Work with partners to build a better delivery mechanism. Bidding on the top 25% most viewable inventory won’t get you 75% viewable impressions. In fact it’s closer to 60% – 65%. If everyone we’re to optimize as we did, it would mean extreme demand on the top 10% most viewable inventory leaving a significant portion of viable inventory underutilized. While there are many pre-bid capabilities already being run in tandem with our digital media, the next step in its evolution is only displaying the ad once it is in view.

What this Means for Digital Media:

We should clarify and further legitimize our industry’s promises. When discussing these buys, we should be able to draw far stronger correlations between media and overall business goals. There would no longer be cheap delivery and inflated impressions. Should, hypothetically, inventory currently average 33% Viewability, to achieve the same revenue CPMs would have to rise 3X for publishers. Overvalued inventory would go down as premier advertisers could display beyond the top 25% or 10% of inventory available while undervalued inventory would generate more revenue. The vast majority of Ad Stacking, Bots and Fraud in their current application would become irrelevant. This would allow for 100% of viewable impressions in all viable inventory, regardless of how the inventory ranks. More importantly, we’d have a true market value of delivery against our target audiences.

jonathanmonroe


J O N A T H A N   M O N R O E
Associate Media Director

 

 

 

 

Posted on February 1, 2017 in Jonathan Monroe

Leave a reply

Your email address will not be published. Required fields are marked *

Back to Top