The consumer of 2017 isn’t tied to a desk when online, and doesn’t shop exclusively in-store or on a desktop. There are more mobile devices than people in the world, and in the U.S., more than one third of the connected population own a computer, a tablet, and a smartphone.
In 2017, the consumer moves seamlessly from mobile web browsing to in-app activity; visits a convenient store; uses a different screen in the car on the way home; and picks up the digital trail again on a desktop or laptop, or on streaming television. Then he or she watches a cable TV show.
For brands, building relationships means not treating the consumer as a complete stranger, never met before, at each individual touchpoint. It means continuing a conversation across multiple screens, controlling frequency of contact with the same individual in different channels, and being intelligently available when and where the consumer wants to be reached.
Ultimately, it means understanding cross device identities.
The Genesis of Cross Device
It all started with mobile phones, and consumers spending more and more time on publishers’ properties in an essentially cookie-free environment. Because advertisers are motivated — obviously — to reach a relevant audience, and because their digital strategy for doing so was built around cookies for targeting and attribution, there was a reluctance to buy ad space on mobile.
Someone needed to figure out how to monetize mobile apps, and the mobile space generally, and the first step would be to build a bridge from environments which supported cookies to those which did not.
It soon became clear, Kate O’Loughlin told me, that fixing user identity across multiple screens created further opportunities — for example:
- More creative personalization
- Better attribution
- Effective frequency capping (ensuring the consumer doesn’t see the same ad too often.
O’Loughlin is SVP of media at the New York cross-device solutions firm Tapad. What’s more, she’s been there almost since day one (she was previously a product director at MediaMath), which means she’s lived and breathed the evolution of cross device identity tracking.
“Now that mobile is eclipsing other platforms for time spent,” she said, “consumers are going to continue to want to have a personalized experience, to have meaningful ads shown to them on their phones, so the need for cross screen is just even further catching momentum.”
Meeting the Challenge
One of a number of early initiatives to solve the cross device puzzle was identifying the devices themselves. BlueCava was one of the pioneers of the device identification approach. Rather than creating cookies, device identification relies on tags inserted in sites to generate signals which it resolves into persistent device IDs it claims are actually more reliable and longer-lasting than cookies.
“The problem with that methodology,” O’Loughlin said, “was that consumers didn’t understand how the identifiers were created and couldn’t opt out of their use.” And the objective, after all, was to not just to collect information about devices, but to resolve back to their users, to generate information about their behavior, interests, and so on (BlueCava now offers a broader portfolio of cross device solutions).
It wasn’t an easy nut to crack. Tapad worked for a time at an exclusively app-based solution, but eventually abandoned it in favor of a more holistic approach. “Our strategy,” O’Loughlin told me, “is to balance scale and accuracy; and so our technique is to blend deterministic and probabilistic approaches into one algorithm that, with confidence, identifies that it’s the same person [on multiple devices] while ensuring that we have lots of coverage.”
“Deterministic” and “probabilistic” are key terms when it comes to understanding how cross-device tracking. Deterministic simply means log in-type credentials, where users essentially identify themselves. Tapad had originally considered using this type of data as foundational, but it provided only a limited framework. “How would you bring in television, or wearables, or other IoT devices?” O’Loughlin asked.
Probabilistic data is richer, but of course less reliable. “You know that it’s probably the same person, using different types of clues,” O’Loughlin explained. For Tapad, these include device proximity, time-based patterns, and context-based patterns:
- Are the devices repeatedly appearing in the same locations?
- Are the users waking up and going to sleep consistently?
- Are the devices being used to look at similar types of content?
The clusters of associations emerge over literally trillions of data points. Deterministic data can be used to verify the probabilistic clusters, bringing accuracy to scale, and generating a perpetually updated Device Graph.
Different Ways to Bake the Cake
Not all cross-device solution vendors take the same approach. One strategy is “house-holding,” based on using the IP address of home networks to tie devices together. That falls short, of course, of tracking mobility. As soon as the consumer is out-of-home, at work or school — or shopping — they’re out of sight.
An exclusive focus on deterministic data is limited too, not least because users don’t always log in. Also, one individual can use multiple log-ins for the same device — which means that from a deterministic point of view, he or she looks like several different people.
But a lighter approach to cross device can work for specific goals. I spoke with Jack Sturn, SVP at 4Cite Marketing, a digital marketing solutions firm with a proprietary identification technology it calls CrossLink. Unlike Tapad, 4Cite does rely heavily on deterministic data, and even cookies. But then 4Cite is focused chiefly on ecommerce and email marketing.
The challenge 4Cite identified was a gap in ESP capabilities when it came to triggering emails. ESPs are good, Sturn told me, at pushing large quantities of emails, but not adept at identifying the online activity which should trigger them. CrossLink can trigger relevant messages in response to website browsing activity, in response to changes in product catalogue, and even in response to a user browsing on competitor websites. It can also trigger website responses, such as targeted lightboxes, when it recognizes the website visitor.
“We pull customers through the tunnel,” Sturn told me, “via repeat emails, lightboxes, and segmentation.” Reliance on deterministic data, he said, “is the simplest way.” It does depend on associating individuals with email addresses, but it’s an approach which might make sense for brands concentrating on linking their website and email marketing efforts (4Cite uses its technology to help manage direct mailing lists too).
Other brands will want to reach more broadly across the digital environment. Cross device identification might have originated in the desire to connect screens, but it’s increasingly relevant to a wide range of digital touch points, including gaming consoles, television (both traditional and app-based, cars, and wearables. “We already have some Teslas in the graph,” O’Loughlin told me. “There’s a browser in the car which appeared in our graph naturally.”
Putting It to Use
Overlaying all this data is what O’Loughlin calls “the application layer.” All the most accurate data in the world brings no value unless it can be activated. There are many benefits to doing so, and they’re much the same for brands and publishers:
- Better insights into customers and prospects
- Better planning tools
- Better execution, and
- Better measurement.
For example, if you’re only looking at an individual’s behavior on a desktop browser (the cookie environment), you only have a partial view. Planning the media that will work for them, and that will maximize reach and optimize frequency, happens more easily when the customer view is comprehensive.
In particular, you can treat established customers like established customers, no matter the channel. “If they already own a particular model of car, for example,” said O’Loughlin, “and you know they’re four years into ownership, then treat them like you know who they are and you have a relationship with them.”
Behind Closed Doors
Of course, the ability to track the same person from computer to phone, and from television to Fitbit, ultimately begins to look like round-the-clock surveillance. What about privacy?
Vendors working in the cross-device space are sensitive to this concern. “We don’t want to just take data from consumers,” said O’Loughlin. “We want them to understand how valuable their data is, and how it’s going to be used. And if they don’t want us to use it, opt out.” The younger digital generation understands that its data is a currency, and, as O’Loughlin puts it, “helps to keep a free Internet.”
Of course, opting out sounds easy. But all too often in a digital environment, the choice isn’t clearly presented to the user. For example, Tapad uses the AdChoices icon, a blue triangle, in every digital ad. Clicking on it tells a user what data has been collected and how to opt out of its use. Tapad said it was the first company to introduce in-app opt outs, and also offers cross device opt out: Opt-out on one of your devices, and you automatically opt-out on all of them.
Of course, demanding action by the user is bound to rely, to some extent, on the user being digitally savvy. The data versus privacy debate isn’t over yet, and indeed Tapad said it isn’t seeing large quantities of opt-outs.
Bringing Offline Data to the Party
If the aim is a comprehensive view of the consumer, then off-device behavior is relevant too. The first step in utilizing sales or CRM data is to onboard it to a cookie which can be used to reach consumers using traditional browsers. A number of vendors, such as LiveRamp, offers this kind of integration into digital. The secret sauce of cross device tracking can then identify that consumer across all their screens. The cookie, essentially, is a first foothold in the process.
Making greater use of existing data depositories is another benefit of the cross device approach. “We were working with an automaker in Detroit,” said O’Loughlin. “They had a huge database of cookies across all of their properties, including third-party relationships” — consumers looking at the product on non-branded websites. They thought they knew their universe.
Looked at from a cross-device perspective, it turned out that a significant percentage of the cookies were overlaps — and the overlaps didn’t necessarily encode consistent data (the same unique individual might be shown as an owner and not an owner in different cookies). Identifying the overlaps meant huge savings in marketing costs, and also enabled better relationships with prospects or customers.
Opportunities with Retail Data
Retail, of course, means very large collections of digital and non-digital customer data: Transactions, website log-ins, mobile browsing, and so on. Using cross device technology, retailers can maximize the value of the data. They can know that a visitor to the store, to the web property, and to the mobile app, is actually one and the same person.
“Imagine that I’m in a Target store one night, and the next day I’m on the Target website,” said O’Loughlin. If the store knows what I bought the previous night, the web experience can be appropriately personalized with relevant recommendations. “For retailers, you can use cross-screen to unify what is going on in brick and mortar and what’s going on at digital point-of-purchases.”
Within retail, there’s relevance for CPG. People do now buy toothpaste and soap, for example, digitally. It’s no longer just a question of digital assisting brick and mortar sales, but each channel supporting the other.
“That’s what consumers want,” O’Loughlin explained. “This kind of fluidity between the virtual and reality.”