• by Mike Caprio , Op-Ed Contributor, March 11, 2016

Like most people in our space, I was surprised to see Facebook’s ad-tech chief Dave Jakubowski announce the end of plans to launch a Facebook multichannel DSP powered by Atlas.  Although Brian Boland, the social network’s VP of ad tech, eventually clarified that Facebook would not completely give up a DSP, but instead focus on mobile, native and video, Jakubowski’s findings from the initial test run were intriguing.

It’s no surprise to people in the know that Atlas has struggled in its attempted renaissance to resume its place as a once-great third-party ad server.  But as one of the largest media platforms in existence, Facebook has done an incredible job applying its technology to place the right content in front of the right people at the right time, and could singlehandedly be responsible for spawning the $100 billion mobile advertising market.  

So what went wrong?

Let’s take a look at each of the reasons Jakubowski pointed out for the Atlas multichannel DSP deprecation.

Fraud/Bots:  The industry has been fighting fraud for many years as nefarious characters from around the world have created new tactics to earn money criminally on the back of our industry.  But there are a number of solutions, including Integral Ad Science, Forensiq, White Ops and others that are designed to protect brands and spend from unsafe or fraudulent traffic -- especially in RTB.

Facebook is no uninformed bidder, so given its vast pool of Device IDs and deterministic data, bots should have been less of an issue.

Low quality inventory in display: In any market there are good quality and poor quality products.  The marketplace typically weeds out poor quality through the process of natural selection.  However, in RTB there is a perceived value, even in low-quality inventory.  By its very definition, an impression that delivers lower value will be priced as such as the market collectively “learns."

Players such as AppNexus have made it their mission to remove low-quality inventory, and buyers create black lists to protect against content, domains or environments they deem as low quality.  This is part of the evolutionary buying process.  

But keep in mind, beauty is in the eye of the beholder.  Not every client is looking at the same KPIs and therefore may value lower-priced, lower-quality inventory to drive reach goals.  This has been the case in TV forever.  Want that hot prime-time show?  Well then, you need to take some day-time fringe to make the CPP/CPMs work.

Not enough real-world value: Offline attribution and ROI has always been a challenge for the digital media industry. Still, we are closer to aligning around that vision than ever before.  Devices, including those from the industry’s two largest players, Apple and Samsung (and Google with Android Pay), have secure payment solutions built in.

In addition, more than 75% of major brands have their own apps, and with location information, foot traffic can be tracked before and after campaign spend. 

For an experienced team like Facebook’s, none of these issues should have been a surprise, including the finding of significantly higher ROI on native and video inventory.  Some industry insiders have speculated that there is just more money to be made in the “ad network” business, but I don’t buy that.  Facebook could “have its cake and eat it too” no differently than Google has.  DBM monetizes AdX traffic as well as outside inventory across exchanges.

I think the more likely reason Facebook abandoned a DSP powered by Atlas that would have enabled the purchase of all ad formats is that the social giant will either buy rather than build, or will leapfrog the market with a new, cutting-edge ad product that gives Facebook more control of the consumer experience than what it experienced in the open market.

Either way, like the rest of the industry, I’ll be watching with keen interest to see what’s next.