Read on MediaPost
By Keith Petri , Op-Ed Contributor, October 24, 2017
Cross-device identity matching is the way marketers map pseudonymous identifiers both intra-device as well as across multiple devices to the same consumer to improve personalization and measurement. It has recently been suggested that cross-device linkages can be improved by starting with the Device ID assigned to an individual.
At a time when the adtech industry is being called to the carpet for transparency, it’s important that we don’t allow claims in a press release to set the agenda. Candidly, cross-device is absolutely not about starting with one ID and not another. Cross-device encapsulates the ideology of clustering any and all (pseudonymous) identifiers belonging to an individual across all of his/her devices – including, but not limited to, desktop computers, smartphones, tablets, smart cars, connected TVs, consoles, and various other internet-enabled touchpoints.
The suggestion that when creating cross-device profiles, one should start with a Device ID as compared to a cookie is flawed and myopic. The argument that cookies are not people-based is also incorrect and the same logic applies to Device IDs. The broader implication that the current industry methodology is wrong and that by focusing on Device IDs instead of cookies, marketers can track actual people at scale, is also off base. The truth is that Device IDs are no more persons based than cookies and both cookies and Device IDs both have inherent limitations. While we must address the limitations for both, we need to agree that the sum of the whole is greater than its parts.
When thinking about cross-device graphs of people, it is important to keep in mind that cookies and Device IDs are just one component of a profile and neither, on its own, will result in true person-based marketing. Both are ultimately perishable wherein Device IDs last an average of 9 to 12 months, but perish completely when the consumer buys a new phone. The lifetime value of a cookie is significantly shorter, however, besides being less persistent, online browsing doesn’t really provide a comprehensive view of who someone is – just what they have looked at online.
I agree that activating cookie-based audiences is a challenge because every ad platform uses its own cookie ID, which necessitates cookie syncing between systems that result in cookie loss. To address the limitations of cookies, a consortium has been formed between LiveRamp and six other ad tech companies who have organized around an effort to build what they are calling a universal cookie. But even if every player in the ecosystem agreed to participate, a universal cookie wouldn’t overcome common issues such as cookie decay.
A standalone Device ID also has its limitations in that it assumes that a user has only one digital ID, when in fact the average person has more than 4 mobile or connected devices each with its own ID. There are also multiple IDs on a single device – web cookies on a device or tablet are siloed between apps – meaning even a single cookie pool has multiple cookies on the same device. Focusing on Device IDs does not solve for intra-device connectivity nor does it connect a single person across multiple devices.
To see the big picture within a cross-device ID graph, one needs to look at a group of IDs that are associated with one another at either the individual level or the household level, and to organize that group into a cluster. Rather than looking at what the cluster is based on – Device ID, IP address, household, cookie, etc., what we should be concerned with is the precision of the cluster; the percentage of identities in the device graph that are truly linked to the same individual. Ultimately, the question is does the solution support cross-device business use-cases, such as multi-touch attribution.
Cross-device is not something that can be just thrown together and it takes expertise and a broad view of the market to truly understand the consumer.