At the beginning of 2015, 90% of U.S. households boasted three or more Internet-connected devices according to Ericsson, with an average of 5.2 devices per house.
These are households we’re talking about, not specific people with their own devices. However, Cisco predicts that by 2017 the majority of U.S. individuals will have five Internet-connected devices on average. Is it that far-fetched to imagine a nuclear home where parents and children all have their own primary laptops, tablets, smartphones, and OTT devices?
Oh yeah, let’s not forget about wearables – iWatches and Pebbles for everyone! Next up, Internet-connected mobiles for newborns that stream Blue’s Clues on command!
While the rise of connected devices has empowered consumers to indulge in a variety of content wherever and whenever, content providers have been left nearly blind when it comes to understanding their audiences. The champion tracker of the Internet, HTML cookie be thy name, is virtually useless outside the desktop.
As non-desktop traffic continues to rise, digital publishers of all sorts are losing their grip on audience behavior… and their ability to effectively monetize it. How are publishers supposed to sell against their valuable audiences that they don’t understand?
Fear not! The tools to retake control are out there. Taking advantage of them requires cross-screen audience strategizing, nuanced management, and an overall comprehensive data strategy. Tracking users on their various devices and understanding cross-screen behavior is more than dropping a cookie or embracing a user-ID system, but it’s not the Sisyphean task it may appear to be.
Answers in Analytics
By now you should have noticed – from traffic analysis and your personal experience – that user behavior differs dramatically from device to device. For a sports publisher, users may quickly check the latest scores on their smartphones, use tablets to enjoy highlight reels or longer-form text content, and then stream games on desktop or OTT devices.
Understanding what your users are doing/consuming on which device is the first step in understanding cross-screen audience behavior. Tying content and where it’s being consumed is also essential to developing a comprehensive data strategy, which needs to be in place before you go trying to track users cross-screen.
A data strategy doesn’t have to be for a set period of time; on the contrary, it should be an ongoing process that is regularly updated based on your findings. At a higher level , you should be deciding what data sets you want to capture – what content-based, audience-based, action-based, and combinations. Next on the map: how do you capture these data sets?
The content segment can be achieved through using analytics tools to understand the traffic flow to and from your website. While there’s a lot to be learned from basic and free analytics offerings – number of pageviews, outbound traffic, geolocation based on IP address – more advanced (read: not free) services can layer in audience features while delivering highly specific metrics.
However, too much data can easily become overwhelming. Publishers should start with evaluating the two or three most important factors. For example: where are people coming from (referral source); where are they going on-site; how long do users stick around; where do they go next?
To bring it full circle, you must then factor in how these traffic patterns differ on web, mobile, and tablet. Then assembling user profiles requires knowledge of how specific users employ devices together, which can be discerned through login IDs.
Y’All Log In Now
Because data is stored differently across devices, the best way to draw conclusions about user behavior in a multi-device environment is to gather data from all sources before stitching them together through a common ID. As we’ve suggested elsewhere, registration data is somewhere close to the sublime. The most straightforward ID for data-sewing is a publisher’s native login.
Not all of your users have to login, but you can study the behavior of those that do on different devices to develop patterns of behavior for audience segments. It’s akin to audience modeling based on the tangible data you have, similar to the process used in audience extension. Modeling also avoids privacy issues by targeting profile characteristics rather than actual users.
But you are going to need a certain level of logins to really drive insights, which will differ from publisher to publisher depending on overall audience size. The key to reaching scale may be incentivizing logins. People will share information if they believe the exchange you are offering is worthwhile. The login system itself can be as simple as employing a service-based login, which is like an automated info retrieval system.
The challenge is determining how much is too much to ask from a social login. This is why a data strategy, and knowing in advance what pieces are most important to your goals, is so important. It will help you decide just what you need from your users and what utility(ies) make for good reimbursement.
“If you keep it as simple as say, email address and zipcode, more of your audience will be willing to use service-based IDs,” says Ashwin Nayak, Quaero VP of Platform Delivery.
Incentivized logins can be simple as adding comment sections or offering customized site recommendations. Basically, you want to encourage users to engage with your content and other users, and the simplest way to do that is by providing smarter, thought-provoking content. But Gavin, these are content problems! Well, talk about how the collapse of the church-state barrier between editorial and advertising has been rampant across the digital mediascape the past few years. Just like viewability, incentivizing logins is another spot where ops and editorial should be in discourse.
Alas, perhaps your site is one of those that simply grab registration data, or you’re struggling to get it to scale. Dry those cryin’ eyes, partner – there are more options out there.
Probabilistic and Service-Based IDs
In the Internet age, everything we do gives off data. Why, we may soon be talking about data in terms of zettabytes, one of which is 1,000 exabytes or a billion terabytes. For example, when an Internet browser visits a site, it shares a huge number of aspects with the web and ad servers. These data points can be as seemingly innocuous as browser type and last update.
The name probabilistic comes from the fact that matching users through such means never comes that close to 100% accuracy – they aren’t nearly as precise as logins. This is both a blessing and a curse: why use a less-than-accurate identifier, especially when login data is available? But probabilistic IDs’ imprecise nature also shields them from many privacy concerns. In addition, users can opt of tracking through services like TRUSTe.
Even though they will never be 100% accurate, providers are constantly improving their data collection and algorithms through experience to raise the bar above 90%. Companies like Quaero use Tapad’s data and proprietary technology to track users across devices and analyze behavior to optimize publishers targeted advertising efforts.
Also, soaking up data no matter where they come from are Internet-based service providers like Google and Facebook, which have developed IDs to tether data from disparate sources. As tools for analysis and targeting, Nayak suggests that these IDs are constantly improving and expects a giant uptick in publisher use this year. Although they can be powerful tools, publishers should be wary of data leakage; keep a keen eye on what information is being shared.
Even more effective are combinations of service-based IDs and native logins – publishers should strongly consider employing options for users. However, the “more IDs the merrier” is not a mantra.
“I’ve seen some publishers using Google, Facebook, Twitter, Yahoo, LinkedIn, OpenID – all that stuff,” Nayak says. “That just gets too confusing.” He suggests that two service-based IDs and one native login make for an effective combo.
The Devices Keep Coming
Those of you flummoxed by fragmentation now are in for some bad news. Particularly as broadcasters and MSOs launch IP-streaming TV services, we’ll be discussing challenges in tracking across over-the-top devices (think Apple TV, Roku, Chromecast, etc.) and gaming consoles. An especially interesting area will be content created specifically for OTT, such as Sony’s Playstation original, “Powers.” In addition, legacy print publishers are increasingly building up video and other interactive content that can be viewed over OTT – for example, WSJ Live.
What about wearables, a trend that has just begun to ramp up? Apps on wearable devices will have fascinating data for advertising purposes, but if you think a mobile screen is small for creative, have you seen a watch screen? The question will be how to incorporate wearables into other channels (and how to deal with a whole new round of privacy concerns).
The “Internet of Things” is no buzzword – connected devices are appearing everywhere, and digital media companies will be constantly re-evaluating where and how to best message their users, as well as monetize them. Essential here is having a comprehensive data strategy, an inclusive one that won’t fall apart as devices and IDs keep entering the fray. As they keep on coming, you’ll be able to strategize with a level head above the hype and buzz.