With marketers constantly struggling to reach audiences across devices, it is becoming increasingly important to understand the ins and outs of cross-screen advertising. While most marketers know that device-matching is an important component of cross-screen advertising, the techniques used to pair devices each have benefits, risks and tradeoffs that should be more widely understood. Deterministic device matching relies on personally identifiable information made available through a login system such as Google, Facebook, or Twitter, or by stitching together data from multiple publishers to make device connections, such as Millennial/Jumptap. This approach allows advertisers to know exactly who the user is and what they are doing, as long as the user is logged in or has shared personal information with the publisher. Pros of deterministic cross-device pairing: This user-based targeting is valuable, as individuals are known with 95+% certainty, and their interests and habits can be monitored across devices and targeted for specific ads. For example, logging into a Google account on a smartphone and in a desktop environment are separate events that serve as an identifiable link between the two devices. A logged-in user searching for flights on their desktop could later see travel-related ads in their mobile browser. Cons of deterministic cross-device pairing: Privacy concerns are by far the largest drawbacks to this approach. The nagging question is whether it’s acceptable -- to users or to regulators -- for advertisers to follow individuals across devices using personal information, such as usernames or addresses. Scaling is also an issue with deterministic device pairing. Many of the deterministic approaches are walled gardens that do not communicate across browser types, applications or devices; ads can only be served in limited environments. For example, Google likely finds it difficult to serve targeted ads on iOS devices, where iPhone and iPad users typically use the default Safari browser, versus Google’s Chrome browser. Probabilistic modeling is used commonly across various fields where large amounts of data need to be statistically analyzed, including in meteorology for forecasting the weather, in pharmaceutical research for understanding drug behavior, and even on Wall Street for predicting market trends. Probabilistic device pairing uses similar algorithms and learning systems to make the prediction indicating, “This smartphone and this tablet most likely belong to the same person.” These predictions are based on observing thousands of data points and attributes from devices -- including device type, operating system, bid requests, and time of day -- and rarely rely on cookies due the disconnect of cookies on mobile devices. These systems, developed by companies such as Drawbridge, Tapad and Adelphic, recognize patterns and make predictions that become stronger over time. For example, if someone frequently visits the same Web page from different devices at the same location, it is reasonable to assume that those devices belong to the same person. Pros of probabilistic cross-device pairing: Because probabilistic device matching does not rely on logins or personal data, the data is kept private and secure. Another positive attribute is that users can be paired across platforms, operating systems and applications. This creates a level of scale beyond even what the largest deterministic platforms can provide. Advertisers can reach users no matter the hot trend in social media or emerging device preference. Cons of probabilistic cross-device pairing: Probabilistic device matching is exactly that -- it’s a probability. However, these algorithms are self-learning, and as they become smarter, the probability that reportedly paired devices belong to the same user will continue to climb higher and higher. We’ve all seen campaigns running across multiple screens for years, but now we’re seeing an increased focus on targeting users, and of course attributing conversions, across devices. As global advertising spending grows, and mobile advertising reaches $41.9B by 2017, marketers will depend more and more on advanced technologies like device pairing to get the most out of their mobile ad spend. Regardless of whether a deterministic or probabilistic approach to device-matching is taken, any marketer running cross-screen ad campaigns should be aware of the ways in which the pairing is conducted, and even be prepared to respond to public inquiry about that process. Transparency and technology have been major topics of conversation in our space for decades, and the rise of cross-device targeting is the intersection of these two subjects.