Recently, there have been a number of public discussions about the future of TV (both linear and VOD) and addressability and measurement in the OTT space. A key component of these discussions is that cross-device addressability within OTT is easy to do and essentially a solved issue and that what the industry is really lacking is consensus and standards. The truth is, cross-device addressability, which is core to OTT measurement, is not solved. Not even close.
OTT addressability which is tied directly to cross-device media buying and identifying audiences within the home and TV environment is still a highly complex issue. While it’s easy for vendors to go into agencies and say that they do cross-device or check the cross-device box, that doesn’t mean that they do it right or even well.
What’s missing from the conversations about OTT is the true complexity of audience identification at a technological level. For starters, OTT environments each have their own proprietary language, the notion of device and/or user ID for delivering ads. And today, the industry lacks a common identifier for use across the various apps on a single device let alone multiple operating systems.
Within OTT there are also category-specific complexities that need to be considered. There are hardware IDs, as well as the application IDs within these devices. About 95% of OTT devices’ interactions with servers take place within a cookie-less environment. So if you’re a vendor that typically works off of a cookie sync and data which is gathered from a pixel deployment for your algorithm, you’re accommodating this new channel by oversimplifying the methodology and relying heavily on IP matching, which isn’t an effective solution.
The truth of the matter is that IP addresses are not identifiers. An IP address is purely an address to or from which internet traffic is routed. The address may represent a single device, a router, or even a cell tower communication channel. And like interactions to and from my mobile phone, the IP address for OTT content is dynamic and many people and devices may be communicating over it, even simultaneously. Therefore, IP alone cannot be used for identification of an individual or even a household.
Relying on deterministic data to associate connected TV IDs to other pseudonymous identifiers simply does not work, because there is significantly limited scale, plus connected device logins are shared at the account holder level. This means a CRM on-boarded ad targeted at Keith Petri would show my mother, father, and sister ads for high-end watches when they are not actually in that audience; they just use my login for their account.
To address OTT, and identity management itself requires probabilistic matching based on behavioral pattern analysis. In a world of OTT, this approach is more valuable, giving marketers the space to define, identify, and target a pipeline of engaged consumers. The same thing applies to the Internet of Things, Tesla cars, and streaming audio. In all of these cases, marketers should strive to understand the linkage between devices in order to better understand their audience and the interplay between desktop, mobile, and OTT. It’s not easy, and it’s not solved.
The OTT space is in fact extremely challenging to solve and as consumer habits change and as TV is redefined as digital video, we need to look closely at how we are identifying audiences in the pursuit of measurement and standards that will make OTT viable for the future.