The Reason Facebook Is Buying Atlas
Given the likelihood of Facebook buying Microsoft’s Atlas platform, lots of speculation has swirled around what the social network company would want to do with the ad server. The popular answer: Facebook is looking to take Atlas’ capabilities at operating across the Web, combine them with Facebook social graph data, create a new ad network and go after the third-party ad serving business. In other words, the argument goes, Facebook is looking to go head to head against Google/DoubleClick.
That might be part of the play here. But I hardly think that’s the whole answer.
To my mind, an Atlas play would really be about attribution.
After all, it is attribution that’s the biggest driver of ad dollars online since day one. Google is a perfect example here: It’s incredibly easy to draw a straight line from the keyword you advertise on, to the text you deliver to a searcher, to the landing page you send the searcher to — and finally to the sale (or failure to make a sale). It’s that ease of attribution that made Google into the most successful direct marketing property in history.
As of now, Facebook isn’t a serious contender for those direct marketing dollars. Just think about how often you’ve heard or read the question, “Do Facebook ads work?” (most common answer: Maybe). If Facebook wants to compete at Google levels for ad dollars, it needs to develop a way to let advertisers and agencies draw that direct line between Facebook ads and ROI.
That’s especially true since digital media has made the entire marketing world far more accountability-focused — to the point that some of the world’s largest agencies are shifting to pay-per-performance models. If accountability were a huge leg up for helping Google in the dot-com era, it’s likely a make or break for Facebook today.
Which brings us to the speculation about Atlas. The ad network theory makes sense because Atlas can 1) do complicated computations around which ads are served when, and 2) draw complex pictures around how users engage across different types of media — including interactions between Facebook and other media properties. Meanwhile, Facebook’s privacy changes from last May allow the site to leverage user data to serve ads to users while they’re not on Facebook.
If you put those points together, it does seem likely that an ad network may well be in the works, and Atlas may become a key piece of technology behind it. But if that’s the only piece of thinking going on here, then Facebook is essentially looking to solve the revenue hurdles of its core business by using that core business to support a side project. That’s an awfully convoluted and less strategic path to growth.
It seems a lot more likely that Facebook is less interested in using Atlas to port data out onto the Web, than it is in using Atlas to bring all those off-Facebook conversions back into the world of Facebook itself. In other words, Facebook wouldn’t look to Atlas as the backbone of a new ad network; Facebook would look to Atlas as the infrastructure for tools that show how users engage with Facebook ads, and where Facebook fits in the conversion path. Those are the tools that Facebook needs to prove that yes, Facebook ads do work. And no matter how lucrative an ad serving business or ad network would be, building those tools is infinitely more valuable.
Of course, beating Google at the ad network game may be one of Facebook’s goals (and it’s certainly something Sheryl Sandberg has the chops for). But if Facebook isn’t thinking first and foremost about its attribution problem, then it’ll never beat Google for ad dollars — ad network or no. Zuckerberg, Sandberg and company are smart enough to realize this. And if they are looking to buy Atlas, attribution is the reason.