In this article you will learn about:
- The type of data sources available post-cookies
- The capability of each type of data source
- Targeting the right audiences without cookies
- The subtle differences between Machine Learning and Deep Learning technologies and which is likely to outperform post-cookie
Today’s model of selecting audiences, primarily based on data brokers, raises many questions regarding the transparency of obtaining information about the user, and the coming changes will significantly impact the data brokerage industry.
In the new reality – starting late 2023 – “target groups” selection will be based on four types of data sources of clear origin. Each of them is expected to have a different scale and scope of information:
- Contextual data
- Behavioral data
- Identity providers
Contextual data means targeting based on the consumed content of pages/articles. It is a deep analysis of websites based on their content’s theme, language, and other data to determine their primary category. All of which can be combined with Brand Safety mechanisms to continue their function.
Behavioral data is based on the idea that the ad should be relevant not to the page, but to the user visiting the page. Behavioral targeting has been widely used in online advertising and marketing for over a decade now and the first party data of advertisers and publishers will still be available for use.
Identity providers are methods based on 1:1 user identification by using a unique identifier derived from personally identifiable information, such as email addresses.
FLoC (Federated Learning of Cohorts) is a brand new approach developed within the Privacy Sandbox. It is based on the idea that the browser groups people with similar web browsing behavior and assigns them to a cohort that can be targeted by the advertising services providers.
You can learn more about FLoC from the infographic below.
Clearly, the more powerful the algorithms are, the more precise the evaluation of the relevant cohort is while bidding for the right audience. The algorithms of different vendors will select different FLoCs, which may cause wide differences in the results.
Thus, using Machine Learning and Deep Learning technologies will no longer be just a technological advantage, but will become an indispensable element in the daily work of every marketer.
Machine Learning is the study of computer algorithms that improve automatically through experience and data use. It is a part of artificial intelligence.
In contrast, Deep Learning is part of a broader family of machine-learning methods based on artificial neural networks with representation learning. This one can be unsupervised by a human to work, so the industry assumes it will be more tailored for post-cookie.
Only the combination of all these methods will allow for a comprehensive approach to advertising campaigns. By combining the different scopes of the delivered data, an accurate new way of flexibly and effectively selecting target groups can be achieved.
If you have any questions or comments regarding the article, please let us know.