Facebook closes Atlas’s video ad-serving gap, widens sales measurement
Facebook’s Atlas can now serve video ads across desktop and mobile and measure how ads lead people to purchase on different devices or in physical stores.
Facebook’s ad server, Atlas, recently came under fire for being less of an ad server and more of a measurement tool, as documented in a recent AdExchanger article. But Facebook is cool with that criticism.
“I actually don’t see that as a criticism,” said Facebook’s VP of advertising technology, Brian Boland. “When you look at what we’re focusing on doing, there’s two pieces. There’s that understanding value or measuring value piece and there’s that delivering value piece. And one of the biggest problems that we’re hearing from marketers isn’t ‘Hey, I can’t serve an ad;’ it’s ‘I’m potentially wasting millions of dollars or spend. I’m potentially missing 40 percent of the conversions that are happening because they’re coming from mobile.’”
It helps that Facebook has shored up one of Atlas’s weaknesses in delivering value — specifically in delivering video ads across desktop and mobile — while building on its strength in measuring that value, i.e., what brands are getting in return for their ad dollars.
Update: At the same time as Facebook is adding to its Atlas ad server, it’s trimming other parts of its ad-tech stack. The company has killed the demand-side platform it tested last year that plugged into exchanges in order to automatically buy ads outside of Facebook through real-time auctions. The decision to shut down the DSP was made because too much of the inventory being sold through the exchanges it accessed was either fraudulent or served to bots instead of actual people, according to a Facebook spokesman.
Facebook’s Atlas can now serve video ads across mobile apps, as well as the desktop web. Previously, Facebook relied on other companies like Innovid to serve video ads for clients but wasn’t able to do so across different devices. The video ad-serving will only work for in-stream video ads, like the pre-roll spots that play before a video, and not the in-banner types that appear within a boxy display ad.
At the same time as Facebook closes the gap on its ad-serving capabilities, it’s widening the breadth of its ad measurement. “Our approach isn’t first and foremost to be an ad server and, oh by the way, we’ll measure. Our approach is to very much solve the measurement problem and to add more value, once we understand the value, through ad serving,” Mr. Boland said.
Brands using Atlas can now track how many people who saw an Atlas-delivered ad ended up buying a product in store, something Facebook has been testing since last summer. To do this offline attribution, a brand would upload a list containing data about the transactions conducted in its stores, such as the names attached to credit-card transactions or email addresses connected to loyalty cards. The information from those lists will be aggregated and anonymized — similar to how Facebook’s Custom Audiences ad-targeting tool works — and then matched against Facebook’s own list of people who saw the ad, which Facebook compiles by being able to recognize if those people are Facebook users.
“We’re enabling the mobile-first world to be observed in the real world,” Mr. Boland said.
The offline conversion measurement will be easiest for brands that own the point of sale, like brick-and-mortar retailers and auto dealerships, but other brands, such as consumer packaged good companies, will be able to bring in sales data collected by a third-party company to use the feature, Mr. Boland said. He added that cash-only transactions where the customer doesn’t provide any information to the merchant, like swiping a loyalty card, cannot be measured.
In addition to being able to see how Atlas-served ads impact real-world sales, brands will also now be able to see how ads served on different devices lead to a purchase or account registration conducted on one of those devices. Facebook won’t be reporting this cross-device path-to-conversion map on an individual level, but instead will present it in aggregate so that a brand can see, for example, if most of its eventual desktop customers had initially seen its ads on their smartphones. Facebook’s hope is that this trend data will show advertisers that even if people are less likely to buy a product on mobile, that doesn’t mean brands’ ad dollars are wasted on the smaller screen. For example, KLM Royal Dutch Airlines used this feature to see that, even though only five percent of the people who convert as a result of seeing its ads did so on mobile, its mobile ads had a hand in 25 percent of digital purchases, according to Facebook.