Consumers now spend more than 50% of their digital screen time on mobile devices, yet you couldn’t tell by looking at the industry’s revenue figures.
In 2014, desktop advertising was a $37 billion business in the U.S., while mobile generated about a third of that, according to the IAB (PDF). Reasons abound for the gap. Smaller screens make it hard to squeeze in a lot of ads. The ad industry is more comfortable buying desktop ads. However, perhaps the biggest factor holding mobile ad spending back is the industry’s relatively slow response to programmatic buying on mobile.
The lack of a standard tracking mechanism on mobile seems to be the biggest barrier to further adoption. In an Ovum survey in 2014 (PDF), 21% of marketers said they weren’t sure whether buying programmatic on mobile was effective, while 26% said they found it confusing.
“If this uncertainty could be dispelled then we think more marketers would move to programmatic buying,” the report stated.
A Cookieless Frontier
The main element that makes mobile programmatic buys so complex compared to desktop is the lack of cookies. While cookies–lines of code that can be used to track a consumer’s movements around the Web–are the standard method used for tracking and targeting on the desktop, they are largely ineffective on mobile, where 86% of activity occurs in apps.
Unlike desktop, with a mobile programmatic buy, you have to take into account a wide array of factors, including device information, app information, geosignals, mobile location data, and mobile identifiers, to name a few. Since none has emerged as a standard, executing a programmatic buy on mobile can be befuddling.
“I’ve been hearing from agencies and other players that it’s really limiting spend,” said eMarketer analyst Lauren Fisher, referring to the lack of a cookie equivalent for mobile. Marketers are still trying to figure out how the other data fits with their targeting objectives, she told CMO.com.
“I think the industry’s working, quite frankly, to make sure all that data’s complete, it’s clean, it’s accurate,” Fisher said. "All of that’s a challenge.”
Even if you take programmatic out of the equation, mobile is complicated, added Suman Basetty, director of product management at Adobe (CMO.com’s parent company). “People go from app to browser, browser to app, and there’s no easy way of linking that,” he said. “It makes it more challenging from a tracking perspective.”
Marcus Startzel, chief revenue officer for AppNexus’s agencies and advertisers business unit, said that most marketers find programmatic hard to fathom, never mind programmatic on mobile. The idea behind it, though, is straightforward: Execute a buy to get ads in front of consumers.
“I would say that the way we talk about programmatic, the ecosystem, does create a lot of buzzwords,” Startzel told CMO.com. “But we don’t all have to fly the plane to make it work.”
Alternatives To Cookies
When mobile advertising emerged, the industry offered several alternative solutions to cookies, including device-identifier IDs, statistical IDs, and universal login tracking. Each has its limitations.
As Michael Collins, CEO of ad tech firm Adelphic, explained, device-identifier IDs lack scale, and publishers don’t always make such identifiers available to buyers. Moreover, Google and Apple run proprietary ecosystems; if you rely on Apple’s IDFA, then you’ll only reach iOS-based consumers. Device-identifier IDs are also ineffective, Adobe’s Basetty told CMO.com, because it’s hard to match them up with mobile Web-based cookie IDs. That results in an incomplete picture of a consumer’s activity on mobile.
Statistical IDs, meanwhile, have scale but aren’t completely accurate because they’re based on estimations. Collins told CMO.com he believes that’s a minor issue because statistical IDs actually are fairly accurate. (While Collins doesn’t divulge a figure for Adelphic’s accuracy rate, Nielsen confirmed last December that rival Tapad’s rate is 91.2%.)
Universal login tracking is accurate, but it also lacks scale because it’s controlled by a publisher. That also means buyers lose the ability to evaluate pricing from multiple publishers.
As a result, Collins advocated relying on multiple data sets for programmatic mobile buys. Of course, that adds to buyers’ confusion because of myriad variables.
“There is a lot of data available in any given mobile ad impression, and it’s different depending on how that ad impression is seen,” Startzel said. “I have an iPhone. If I’m on Wi-Fi at a hotel, I look very different than if I’m on LTE or 3G.”
Also adding to the confusion is cross-device targeting, which, proponents argue, let you target people, not their devices. If you execute a buy via Facebook’s Custom Audiences, you could show a desktop user who searched for a certain brand of bike an ad from that company on Facebook’s mobile app. Facebook also uses such data to target ads on third-party sites and apps.
While Google hasn’t launched a cross-device solution for advertisers yet, others, including Yahoo, AOL, and Twitter, offer similar “deterministic” methods of targeting users as they go back and forth between their devices.
For those who balk at a walled-garden approach, several “probabilistic” vendors, including Adelphic, Tapad, Krux, and Drawbridge, pool huge amounts of data from desktop and mobile to make educated guesses about which consumers own which devices. While such solutions often work well, they add to the complexity of an already complicated programmatic transaction.
A Mobile Standard?
A universal standard tracking tool for mobile–preferably one that included desktop data, too–would be one way to simplify mobile programmatic buys. While some believe such a standard might emerge in a few years, none has yet.
In 2013, USA Today reported that Google was working on an anonymous identifier for mobile and desktop called AdID, but Google never confirmed the report. Even if Google released such a solution, though, it would hardly be universal, but would likely be available via Android and the Chrome browser. (A Google rep declined comment on the 2-year-old report.)
Statistical IDs, which are based on data such as IP addresses, browsing, and device type, among other factors, so far seem to be the only alternative. There’s no statistical ID standard, though, since they’re all proprietary solutions.
Since marketers are more interested in targeting users on mobile than delving into all facets of the technology, the onus is on middlemen to make it easier for marketers to execute buys.
“I think the industry and the platforms on the supply and demand side are removing a lot of that complexity and bringing to the surface the proper controls for a buyer to be able to find and engage their audience in a straightforward manner,” Adelphic’s Collins said. “The new tools strip out a lot of complexity and pulls that to the background.”
Even if programmatic on mobile remains challenging, there’s enough momentum for both mobile advertising and programmatic to power through. An October eMarketer report predicts that mobile programmatic will be a $9.3 billion business in the U.S. in 2015–with about $5.9 billion coming from Facebook. Then the market will rocket to $14.9 billion in 2016, when it will eclipse programmatic on desktop by a wide margin.
“It’s going to get easier for marketers to use these solutions,” eMarketer’s Fisher said.
See what the Twitterverse is saying about programmatic advertising: