Can Snap make an Audience Network work?

Snap, the company that publishes the Snapchat app, held a partner summit last Thursday at which it unveiled a number of fairly substantial new product and content features. The announcements included a lineup of eight new original content series, a new AR implementation called Landmarkers, an in-app games platform, a tool for bringing Snapchat-style stories to 3rd-party apps called Story Kit, and the Snapchat Audience Network, an ad network that will allow Snap advertisers to reach users in 3rd-party apps. Snap’s stock price surged on news of these initiatives and now trades at nearly two times its price from the end of 2018.

Given its name, it’s impossible to not immediately draw comparisons between Facebook’s Audience Network, which launched in 2014, and the new Snapchat Audience Network. When Facebook’s Audience Network (FAN) launched, I wrote:

But what’s more, in building out an ad network, Facebook can tightly integrate its own social graph into the fundamental mechanics of mobile distribution, strengthening its position against the broad threat of “unbundling” on mobile. Just as Millennial Media acquired Jumptap and Twitter acquired MoPub in land grabs for valuable mobile user profiles and data collection infrastructure, Facebook’s new mobile ad network will benefit immensely from the data its social graph can lend to targeting. And the more developers that integrate Facebook’s API into their apps for advertising, the more data the company will accumulate.

The key here is that in launching the Audience Network, Facebook was allowing advertisers to reach Facebook users — which effectively means every smartphone owner in the developed world at this point — in any app in which they engage and not just in Facebook-owned properties. This is a powerful idea not only because the Facebook news feed was always destined to reach a maximum ad load, but because Facebook inventory is highly prized and competed over, making it expensive. If Facebook could allow an advertiser to target users using its own proprietary profile models and find those users in apps that aren’t competitive in terms of impression costs, then it could allow advertisers to target users cheaply while still remaining dependent on Facebook’s data.

This was a beautiful transposition of the Facebook value proposition, turning a source-level targeting schema (“I advertise on Facebook because the ROI performance of that source is favorable”) into a user-level scheme (“I target certain Facebook users, wherever they are and using Facebook data, because the ROI of those users is favorable”). The Facebook Audience Network effectively extended the advertising surface area of Facebook’s proprietary user data to the entirety of the mobile app economy: by joining its own proprietary profile data to the advertising and device IDs that it sees for its own users, it creates opportunities for advertisers to reach those users in all (or at least, more) of the apps they use. The diagrams below are meant to try to illustrate this dynamic:

Of course, Facebook has more data at its disposal than just advertising IDs to use to join data on: it can use a multitude of data points to identify users, whether they have Facebook accounts or not (some details of these activities were surfaced in Mark Zuckerberg’s Congressional testimony last year). But even if Facebook could only identify Facebook users in FAN participant apps, that still represents fairly significant progress towards identifying every person on Earth: Facebook reported 1.52BN daily active users for Q4 2018.

And that fact raises an interesting question over the viability of Snap’s version of the Audience Network: can it work with an order of magnitude less user scale? Snapchat’s latest reported DAU metric was 186MM; can that provide the level of scale needed to grow a meaningful 3rd-party advertising business through the Audience Network?

But beyond that, some other questions call the viability of Snap’s endeavor into question. First, Snap doesn’t have nearly the level of granular detail around interests and behaviors that Facebook does. How valuable is Snap’s data in targeting users, on Snapchat or off? Brand advertisers may be happy to shower Snap with their campaign dollars to reach its young audience, but Snap has struggled to prove to performance advertisers that it can deliver ROI-based efficiency. If Snap can’t deliver that for its first party inventory, can it do so for third-party inventory?

And the second question that remains unanswered is how ad creative will be handled for the Snapchat Audience Network. In the unveiling last Thursday, Snap emphasized the fact that this audience network would bring its signature full-screen video ad format to third parties, but there are some problems with that. First, not all apps — and especially many games, which represent a massive vertical slice of the mobile advertising market — run in a vertical screen orientation. Second, that format may work well within Snapchat, where users have grown accustomed to it, but it may underperform relative to other creative concepts and become a liability out in the wild. With the mobile advertising ecosystem becoming increasingly programmatic, rapid creative experimentation and flexibility is a competitive advantage.

Snap showcased some incredibly interesting product initiatives at its partner summit last week, and it does seem to be making forward progress with innovative new engagement mechanics. But it’s tough to beat Facebook on its own turf in a game that it largely invented: performance mobile advertising. And while many advertisers would love to see new contenders insert themselves into a marketplace that is wholly owned by Google and Facebook, Snap faces some formidable, systemic headwinds in gaining traction with its audience network.