It all comes back to Identity

AI is the new buzzword. AI is going to do everything, like some magical elf running around behind the curtain automating replicating human critical thought.

But AI is only as powerful as the data it has access to.

Growing up, my father used to say, “a half-truth is a lie.” It’s a lesson I never forgot, and something I constantly think about. A point of data can be accurate, but missing context, giving an incorrect overall picture.

Which is why Identity in data is so pivotal. Identity is how we join the disparate collected data across disparate systems to put together a comprehensively accurate view of a journey. This is in stark contrast with the push to privacy regulation and industry changes.

Silo’d Measurement

With deprecation of IDFAs (effectively), and increased deprecation of 3rd party cookies (Safari, Mozilla)…seeing a comprehensive journey across ad platforms is increasingly difficult. The effect of this is multiple different ad platforms, within silo’d tracking eco-systems, taking credit for the same conversion.

Talk CPA’s or CPIA’s all you want, but I want to see the underlying data and how you’re reconciling duplicate conversion attribution.

An identity strategy adapting, and pivoting, to the changes in industry privacy and regulatory standards is critical in effectively measuring your advertising spend.

Limits of 1st Party Data

First party data is the (rightful) golden standard of data. Data collected directly through your 1 to 1 relationship with your consumer. But there’s an implicit hole in it – it can only tell you how a consumer behaves while interacting with you, not when they’re interacting with others.

This is where third-party data becomes critical to a comprehensively accurate view.

Models built solely on first party data are biased towards the guard rails of the first-party relationship. In more practical terms, in the streaming world, building models solely on content a consumer watches on your platform is biased by the type of content you offer. Let’s say you offer a lot of drama, many first-party modeled consumers will appear as high-drama genre affinity viewers – because you don’t see that they spend double their time watching sci-fi on another provider.

This is the very problem the buzzword’ Clean Rooms aim to solve. By anonymizing identity, non-identifiable profiles of real consumers join first party data and licensed third party data sets to get more comprehensive views of a consumer’s actual behavior, interests and actions. A look-a-like model is only as good as the attributes it has to compare.

Data Interoperability

Even first party data can be disparate. Different systems collect different stages of the consumer journey, across different devices – that are personal or communal. In the world of streaming you’re dealing with browsers (cookies), mobile apps (MAIDs) and CTV devices (Roku ID, Fire ID, etc.).

Cookies and MAIDs are generally a 1 to 1 relationship with a consumer, while CTV IDs are household IDs and there’s almost no way of knowing which household member is watching content on the CTV device.

These complexities complicate measurement, personalization, modeling and, to go full circle, AI.

Identity is connector between all of this data, whether it be joining disparate systems, devices or connecting 1st to 3rd party data. Yet so often an identity strategy is an after thought when working through a data strategy.

Your data is only as good as your identity.