Produced in association with Digilant
People shop mostly on their desktop computers—but they live on their smartphones. For marketers, effectively reaching their target audiences requires making a connection between those two worlds.
Marketers want to know how effective their ads are on mobile devices. They would also love to be able to retarget a mobile shopper who began making a purchase on that device, and then gave up before providing payment information and a delivery address.
Cross-device tracking is challenging, but technology companies are developing tools to make it easier. Using probabilistic identification methods, they are linking smartphones to desktops accurately enough to justify ad placements, especially given that mobile ad rates are substantially cheaper than desktop display ad rates.
The smartphone environment makes digital marketing difficult. Apple’s Safari browser for smartphones and tablets prevents tracking cookies from being installed, blocking the most common means of detecting a user. Many phone users jump from app to app without using their browsers, which complicates tracking. On top of everything else, “you tend to overcount clicks, because many are caused by fat fingers,” notes Kim Riedell, senior vice president of product and marketing at Digilant, a customized programmatic media solutions company in Boston.
Apple does use an identifier for iPhones that makes it possible for advertisers to track user activities. Google’s Android operating system permits cookies, and Google has also launched its own Advertising ID.
Even in those cases, it’s difficult to follow a user from phone to tablet to desktop unless the user logs in to an advertiser’s page on each device or uses a browser that permits tracking. For apps such as Facebook, where users generally use the same log-in information, cross-platform ad attribution and retargeting is possible, as long as the user clicked from within the app or within the Facebook application.
Recently, data management companies have started using statistical analysis to make educated guesses about user identity. They can hypothesize, based on certain limited pieces of information, that a given smartphone user probably is the same person as a given desktop user. Among the data they collect and analyze is information about the Wi-Fi networks a person uses, the websites she regularly visits, time-of-day patterns, and geographical cues.
Adometry, a data-analytics firm based in Austin, Texas, says it’s able to use third-party data and its own analysis to tie all a user’s devices together through a technique called device-mapping. Adometry uses technology from Tapad, a New York City-based company specializing in cross-device content delivery, to map desktops, tablets, and smartphones to a single user.
Tapad’s probalistic analytics are 80 to 85 percent accurate, depending on the number of data points collected, according to the company’s CEO. Probabilistic analytics look at patterns of usage, geography and the time of day to predict that the person reading The New York Times on her smartphone is the same person reading it on a desktop a few minutes later. Deterministic analytics are even more accurate because they rely on following people who use the same log-in from different devices—but deterministic information isn’t as widely available.
Drawbridge, of San Mateo, California, says it can target more than one billion mobile devices with ads that are also relevant to those seen by owners on their desktops. Founder and CEO Kamakshi Sivaramakrishnan says that her company can “take anonymous signals from the device and do a kind of statistical space-time triangulation.” By performing the analysis over time, Drawbridge identifies clusters of devices and then figures out which are paired, providing confidence that they have the same user. The results provide marketers with data that is accurate enough for retargeting and attribution.
Other companies also hold patents on their own cross-platform technologies. Adelphic Mobile of Waltham, Massachusetts, for one, uses aggregated data about cookies, app usage, and devices to create a unique identifier for a user. However, because these companies all use different proprietary technologies, marketers cannot accurately compare results. Until the industry determines a standard, such comparisons will continue to be difficult.
Adometry’s chief marketing officer, Casey Carey, has argued in The Economist Group’s marketing blog that the industry would be better off with “a universal identification mechanism that stitches the data together across devices, networks and platforms.” Such a mechanism would provide benefits to advertisers, Carey says. It would also help users concerned about privacy create consistent opt-out policies that could extend across their total Internet usage.
At the same time, many of the biggest companies that possess information collection ability see little benefit in sharing with their competitors. So cross-device tracking and targeting is likely to remain an inexact science for the near term.