Between cookie deprecation, browser changes that limit attribution, new data laws and the potential loss of addressable open exchange inventory, it’s clear that data and identity are the most important topics in the ad industry today.
We’re talking about some of the most disruptive changes to digital media in the past 15 years – changes that will permanently alter media operations and data activation. Put simply, the proverbial sky is still falling.
Yet this moment in time is also the perfect opportunity to build a better ecosystem. The old, broken digital media system put brands, media agencies and ad tech companies in the center, leaving consumers on the outside – even though their data powered the whole operation.
If we are to build a more virtuous circle of data usage, consumers must be at the center.
How do we get there? The industry must break its addiction to deterministic data.
Weaning off deterministic data
Breaking this addiction starts with understanding two key points in the digital ad-serving process: the point of impression and the point of measurement. The point of impression is where the consumer is identified and targeted with an ad on a publisher site or app. The point of measurement is everything that happens after the impression is auctioned and served.
The current way of transacting media is to use deterministic targeting at the point of impression and deterministic measurement and attribution at the point of measurement. This does not always reflect sound privacy practices.
The most virtuous future separates the point of impression and the point of measurement, making use of probabilistic data at the point of impression and a combination of deterministic and probabilistic data for measurement and attribution.
Brands may think that, without deterministic audience data, they have no chance of effectively reaching target consumers. It often seems that the industry at large would rather cling to deterministic data, and all of the associated privacy issues, rather than explore new approaches that can scale and perform.
But contextual data and predictive audiences (or cognitive data) are probabilistic data sets that are already widely available within programmatic marketplaces. This insight makes it easy to target a consumer based on inferred interests and anticipated responses – without the use of either deterministic or device-based identifiers.
And neither of these data sets violates the end consumer’s privacy as they are both cookieless and deviceless and not anchored in any personally identifiable information. Better yet, advertisers can use these signals to build out cohorts or vector-based audience targeting that works at scale.
Using probabilistic data intelligently
First, we need to break the false narrative that probabilistic data means fingerprinting. This is a fear-based trope used by marketers who pedal deterministic solutions anchored in PII. All probabilistic data is not fingerprinting.
The two are not synonymous. Contextual and cognitive data are devoid of deterministic identifiers and device-based identifiers, and they are also not anchored in PII the way many deterministic data sets are today.
The future of advertising depends on embracing new models, such as privacy-safe probabilistic data. It also depends on finding new places to apply those data sets to media, such as from the supply side.
The dominant form of targeting today applies third-party cookie data upstream, at the DSP level, when buying open exchange inventory. These segments – whether they are custom-made, ported from a DMP or bought off the shelf in a DSP – will lose 80%-90% of their ability to target open exchange inventory once cookie deprecation becomes a reality.
The future depends on applying cookieless data downstream directly to the inventory at the SSP level. This allows for better targeting of open exchange inventory, as well as private marketplaces fueled by deal IDs.
This approach not only solves many data privacy issues that advertisers are facing, but also creates transparency.
Plus, applying data directly against inventory from the supply side is proven to effectively optimize the supply path, removing some of the middle layers that advertisers are trying to avoid. It also opens the door to optimize the audience and inventory composition of the deal IDs, which isn’t possible in today’s open exchange using DMP data only.
In short, applying probabilistic data from the supply side can provide the addressability, scale and performance that so many advertisers fear they’ll lose. It can also provide the optimization and efficiency that have long been promised by programmatic buying, making this next generation of probabilistic data really valuable.
The question is: Are advertisers ready to give up their addiction?
“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Follow Audigent and AdExchanger on LinkedIn.
For more articles featuring Drew Stein, click here.
