Cookies have historically been used for identity, allowing platform-to-platform communication via a cookie sync and bouncing data off of a browser. With cookie deprecation, brands and their platforms will now have to communicate with first-party data to deliver targeted advertising. Because this data is much more sensitive than the third-party data used with cookies, it will require a combination of direct connections and knowing your counterparty.
Many brands and platforms are in the process of building APIs that should theoretically allow platform-to-platform communication. But as we start to see early results, it’s become clear that building an API alone is not enough. Platforms have to process and prepare their data for queryable access. The objective is to enable a back-and-forth using queryable data sets to allow platforms to communicate and deliver an outcome or solution.
That’s the key word: queryable. While advertising is a data-driven practice, brands and agencies can’t execute their plans with raw data. They need the insights from that data in order to plan, target, and optimize. Data companies and platforms have the data, however it needs to be processed and configured so that it is available “at rest” and on-demand to users who are able to query it.
Users of advertising technology platforms don’t always care how the sausage is made, but with third-party signal loss looming, the inability for on-demand access to data is going to create a very large disruption. Product and tech teams are building these access layers to efficiently open the door for on-demand access of data where it rests.
The cookieless future necessitates that marketers access data, on demand, within the platforms that they are using for the work they are doing at that moment. A buyer using a seat on The Trade Desk to run a campaign may want to do some planning work and determine which publishers they can run the campaign across. In an ideal world, that buyer would be able to query a list of top publishers to find the best set of publishers for reaching the target audience. When it’s time to optimize, the buyer should be able to pull in campaign measurement data to their seat so that they can optimize and make decisions in real time.
Right now, the more likely situation is that the buyer would need to swivel to a seat on a different platform and try to match the data between two systems. This is a process that could take days, weeks, or even months.
To make matters more difficult, there is no rulebook that is applicable for every company to make their data queryable. When one platform pushes a large data file, the receiver of the data has to store and process it to generate the analysis and insights required for use.
To prepare for the future, the data owners will actually need to do that analysis in advance and then allow users to access data that has already been processed into queryable information.
For example, marketers are about to spend a lot more time working to deduplicate users within data sets. If they access a tranche of IDs via an API connection to a clean room, the issue is that no work has been done to link the IDs. In order to create a single profile, the receiver has to analyze it and determine the deduplicated users across all of the device IDs. That’s an extensive (and expensive) graphing exercise.
In essence, we’re talking about turning the data into something queryable, rather than just raw data. Queryable data is already processed and can be sliced, diced, and filtered in real time.
This is what’s required to replicate or improve on the functionality that is delivered via the current third-party infrastructure using first-party data. Ad technology has long promised the ability to use a wealth of data in real-time to forecast campaigns or provide frequency curves. With this shift to first-party and the development of queryable data access layers, there is a clear opportunity to finally deliver on these promises.
Nowhere is this more obvious than in the rapid adoption of AI. AI works best when it receives pre-processed data that is easy to analyze. Feeding raw unstructured data into an AI system will result in inconsistent low-quality insights and output, aka garbage in garbage out.
The driver behind all of this is coming from the current privacy movement which has forced companies to keep data within their walls. The headline-grabbing breaches and leaks are a byproduct of moving data around. That practice introduces security risk, and it’s something that many data companies are tired of. When data providers set up their data to be queryable and accessible at rest, they maintain control over their data.
The future of digital advertising relies on accessing data at rest, rather than massive file transfers. Ad buyers will still be able to access data, query it, capture insight, and pull back the set of audience IDs that fit their segment. The difference is that the only thing moving is the set of hashed IDs, eliminating any link back to the person or human being, while enabling platform-to-platform communication.