Skip to main content

Cross-Device Targeting – The Mythical Unicorn of Digital

by Mona Khaldi
As part of the Marketer’s Guide Series
June 25, 2015

As consumers, we engage with brands without giving much thought to the various screens we experience them through - we may browse a catalog on a tablet, research products in-store on a smartphone, then head back to our desktop to make the final purchase.

In fact, according to eMarketer, in 2014 we spent almost a quarter of our media time on mobile devices - a 66% increase in activity from 2012.

With this divided engagement across screens, being able to identify unique users browsing between devices has become a critical capability for marketers looking to effectively target potential customers at every stage of the path to purchase.

Until now, cookies have been the standard to identify and target users browsing on desktop devices. Cookies, however, are limited for mobile, cannot connect the dots back to desktop, and therefore aren’t effective as an identifier of devices.

The industry challenge has been to find a universal identifier – a point of reference that can recognize anyone, using any device. And here is where the greatest barrier remains - waiting for technology to catch up with the theoretical.

In the context of digital media, cross-device targeting technology is still in its infancy. Thus far, no single company has been able to develop a comprehensive solution, but Facebook, Google, and Apple have built platforms to capitalize on the potential.

So what are the options available to marketers today? Identification and targeting can be done via 3 methods: cookies, probabilistic identification, or deterministic identification. Each method’s differences lie in which data points are used to identify individuals.

Cookie-based targeting is the oldest, most scalable, easily deployable, and yet least accurate method of targeting cross-device. A cookie is a piece of information gathered from a website, that is then stored in the user’s browser and sent back to the server notifying visited websites of the user’s activity.

Cookies are still the go-to for desktop display, but their usefulness has begun to decline as consumers increasingly rely on mobile devices through the sales and conversion cycle.

Next, targeting using probabilistic identification refers to using publically available ad-serving data, such as device, browsing behavior, location, OS, etc. Using a mix of algorithms, these data points are pieced together to predictively identify individuals.

With match accuracy between 50-90%, probabilistic identification is much more accurate than cookies. For marketers, that may reduce wasted budget spend on the audiences they are not trying to reach, but experts have also remarked that probabilistic targeting methods are unclear, lack standardization, and are not scalable to large audiences (Forrester).

Lastly, there is targeting based on deterministic data - this uses known user information such as login detail, email address, and customer ID. So far, only email providers and communication and social networks have access to a substantial set of this data, and not all are scalable for mobile – but, the amount of information gathered creates the potential to capture individual user profiles.

This method of deterministic targeting is gaining traction in the agency space, with recent evidence of Facebook’s re-launch of their Atlas ad server, which uses a mix of cookies and Facebook IDs to target specific users across devices, within - and beyond - the Facebook ecosystem.

While not everyone uses Facebook, this is a significant stride forward as over one sixth of the world’s population holds an account.

Google is challenging this by building Doubleclick Audience Center, a Data Management Platform which will also attempt to match data with identities, as it has access to billions of search engine, Android, Gmail, and YouTube users.

Additionally, data and capability integrations, such as the Verizon acquisition of AOL, are a step in a similar direction as they combine deterministic data - Verizon’s access to email addresses, browsing histories, phone numbers, and physical addresses, with ad tech capabilities - AOL’s programmatic platform. 

While the scalability of the deterministic data approach is limited, logins and persistent IDs are certainly more scalable than past targeting methods, as they hold a much deeper level of audience knowledge. Which brings us to question – is this the direction we should be heading in, as both marketers and consumers? Are persistent IDs the end-all?

A recent report from eMarketer concludes that a universal or all-encompassing solution would likely have to use a “ubiquitous” key to identify users - meaning it would have to be developed as an industry standard with input from both governing bodies and stakeholders across the industry.

In order to meet demands of marketers, it would also have to offer unlimited audience reach, give advertisers and agencies more control of proprietary data, and provide a holistic view of the consumer journey.

Consequently, it is just as important that the solution benefits the targeted audience; better targeting means serving advertising that is relevant to user interests and needs, without invading their privacy.

A study conducted by the Digital Advertising Alliance in October 2014 found that of Americans who expressed an opinion on mobile advertising relevance, a majority of nearly five-to-one preferred seeing ads relevant to their interests.

What cross-device targeting can ultimately do is help marketers enhance the user experience across the full range of their devices and bring customers closer to the brands and messages that fit best.

To learn more about the vast digital space, check out our Field Guide to Digital Marketing!