This week, Google announced that in late 2022, when Chrome begins to block third-party cookies, it will be operating exclusively on Privacy Sandbox’s anonymized audience cohorts. More specifically, Google will not be creating or accepting UIDs for the purposes of targeting or measurement.
For months, the industry has been working on proposals to submit to Google’s Privacy Sandbox and the greater industry via REARC working groups and other associations. Since the announcement of the deprecation of cookies, there have been several viable scenarios presented that would ultimately allow advertisers to measure the effectiveness of digital marketing in a “privacy by design” structure eliminating third-party cookies, fingerprinting or any cross-site tracking. It meant the re-invention of the ecosystem.
Enter Audience Cohorts
Cohorts refer to large groups of people (or “web crowds”) that exhibit similar browsing patterns. The crowds get clustered together into cohorts that reflect audience segments we see today. This framework allows for customization as well. In previous announcements, Google has stated that early testing of the cohorts-based approach would allow for anonymized addressability at scale. The headline take-away is that Google is putting a big bet on the scalability of guaranteeing anonymity to provide a global approach to privacy legislations that may continue to arise as the years go on as opposed to creating potential tech debt against consented audiences and the continuous amendments that may be required to comply in each market.
The trick will be to guarantee scale without sacrificing the granularity of more nuanced audience segments. Now that the direction is confirmed, the work will need to be done around measurement and attribution as well as the viability of cohort sizes that balance the need of privacy and value to advertisers.
Early work suggest that the user groups themselves will replace device-IDs or cookie-based identifiers, to serve as the basis for everything from targeting and re-targeting to campaign management and attribution. There is a lot of math to be done in the coming months.
What it means to Industry Stakeholders
For advertisers, we continue to recommend getting first-party data stores in good order. This means establishing meaningful consent and looking into viable segmentation models that will become the winning formula against competitors. As always, the content strategy comes into play as well as the ability to generate meaningful engagement while capturing accurate intention signals. More best practices to come from our working groups.
Publishers must continue to build out their first-party strategies as well. Incorporating sophisticated contextual signals that are developed to interoperate with advertiser customizations. Again, think about granularity and originality of cohort creation as well as the dynamic movement of signals. Adoption of IAB Tech Lab’s content taxonomy will provide a helpful grounding for future state cohorts.
Ad Tech will also need to lean in to develop systems that work with an innovate against the new audience currencies. We see an enormous amount of involvement at the IAB Tech Lab as well as other associations on this front.
IAB Canada will continue to contribute to and report on the work ahead with Rearc. Two core approaches to addressing audiences will require standards and guidelines as well as resources like DataLabel.org and the Audience Taxonomy project.
As for the way in which media will be bought and sold programmatically, the work of Privacy Sandbox continues and there have been several proposals related to frameworks addressing where data sets will reside (centrally or decentralized etc.). For all of this to work, the cohort data must be able to securely feed into any machine-learning decisioning and modelling, and where the advertising auctions could physically take place. Some proposals outline scenarios where the data would be housed in-device/browser while others describe a certified publisher hosting framework. There is also the possibility of hosting through neutral and independent third-parties.
We are also continuing to explore and present the incredible innovations that are evolving to address contextual and behavioral signals for publishers. There are exciting things happening on this front.
Following are some brief outlines of recent updates and proposals – warning, there are a lot of birds here:
Two Uncorrelated Requests, Then Locally-Executed Decision On Victory – this serves as a building block of the sandbox proposal that addresses separating any interest groups (cohorts) from any contextual data and publisher-provided IDs when executing a client-side programmatic auction.
Federated Learning of Cohorts – FloC proposes an API that extends Chrome and provides access to machine learning algorithms that apply to the behaviors or interests of large numbers of users in order to better develop and enhance the cohorts based upon the user behaviours and sites that an individual visits.
(First Locally-Executed Decision over Groups Experiment) – is a proposed public trial iteration of TURTLEDOVE based upon industry feedback. Scheduled to take place later this year, FLEDGE will provide the AD Tech community to bring their own ad server model into the API for testing, which is dedicated to the on-device approach.
Independent AdTech companies have developed several proposals that will be incorporated into FLEDGE trials.
Here at IAB Canada, our TCF policies are in final days of public commentary and we are currently working with IAB Europe to establish use cases given the new realities we face.
There are a lot of moving parts here. One thing is clear, the industry is making progress and with three approaches coming into greater focus, the tasks at hand are becoming much clearer. Preparing for the three scenarios – authenticated IDs, contextual signalling and anonymized cohorts is imperative – NOW.
Join our working groups and join the conversation. Reach out to firstname.lastname@example.org to lean in.