“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Devin Guan, vice president of engineering at Drawbridge.
In the weeks since Apple previewed iOS 8 in June, there has been a lot of talk in the digital advertising industry regarding one ostensibly small change that Apple announced. Beginning with iOS 8, slated for release this fall, Apple will randomize MAC addresses when scanning for Wi-Fi networks. Naturally, there are a lot of questions about what a MAC address is, and what this means for our industry.
MAC addresses are used as reliable ways to identify devices by many ad tech vendors. Some even leverage MAC addresses to match devices and establish cross-device identity. The initial concern was that without reliable MAC address information, can there be reliable cross-device information?
Though some ad tech vendors may face growing pains as they react to this shift, overall Apple’s decision will be a good change for the industry, as it gives users more control by removing a deterministic feature that some ad tech vendors rely on to establish cross-device identity.
Impact On Advertisers
A recent Google report indicates that 90% of multidevice users admit to switching between devices to accomplish online tasks, so there is no doubt that cross-device identity and targeting is critical to delivering a better advertising experience for users, better campaign metrics for advertisers, and higher inventory value for publishers.
Cross-device identity can be established in two ways. The first is through using deterministic methods, in which user-submitted or fixed data (such as permanent identifiers) helps create a match across devices. Examples of user-submitted or fixed data include social media or other logins, HTML5 local storage, and of course, the MAC addresses in question. By observing these data points and linking the fields across platforms, technology vendors can establish a confident match between a user’s devices. Then those vendors can offer robust advertising tools that practically guarantee that they can reach a specific target audience across those users’ desktops, smartphones and tablets.
Ad tech vendors that use user-submitted or fixed data for their cross-device identity or targeting could be severely affected. Prior to iOS 8, MAC addresses were among these fixed fields that were core to many vendors’ device identification and, by extension, device-matchingtechnologies. Because MAC addresses will no longer be a reliable data source, some vendors will need to rework their device-matching strategies.
The second method is probabilistic identification, or predictive modeling, where enough data points are gathered to make a prediction that multiple devices belong to the same user, as well as the idea of a user’s interests and demographic information. Typically, predictive modeling does not use any user-submitted data, instead relying on identifiers from event logs, such as ad requests, which are stripped of data that can identify a user in real life, and are ingested by machine-learning algorithms to make predictions.
However, ad requests and conversion pings may contain MAC addresses, so even cross-device technology vendors that use probabilistic methods could face problems this fall if their algorithms are reliant on data using this permanent identifier. Their algorithms will have to adapt to the dynamic natures of device IDs, which are similar to browser cookies, and regenerate data associated with them.
Given this move by Apple, and the many user-controllable identifiers that ad tech providers still have access to, cross-device identity providers should be moving towards a model that is not reliant on permanent identifiers. This way, advertisers and publishers will still be able to leverage cross-device technology for better advertising, and users will feel more comfortable knowing that permanent identifiers aren’t being used.