- 31 de enero, 2020

Comscore Measurement in a Cookieless World

Frank Pecjak
Frank Pecjak
Senior Vice President, Innovation

Google’s announcement of their intention to phase out third-party cookie support for Chrome within the next two years is the latest step in the evolving privacy-centric ecosystem. During this two-year period, Google, through the web standards community, is seeking feedback and adoption of APIs and infrastructure for its Privacy Sandbox. The concept is centered around keeping user data within the browser (under user control) and, once permission is granted by the user to share, limiting data released to the minimum scope for performing the stated purpose.

Comscore is in full support of these privacy-by-design principles and looks forward to working with Google and the rest of the community to continue to develop these important tools. Comscore’s methodologies and approaches to audience measurement are well aligned with the principles laid out for the Privacy Sandbox. These methodologies and infrastructures have been a core part of Comscore’s DNA from the beginning of our intake of census data and continue to be a driving force in our roadmap to evolve audience measurement.

Comscore Digital Measurement Methodology
Since the release of Unified Digital Measurement (“UDM”) in 2010, Comscore’s audience measurement philosophy has centered around using the context of our first-party panel data to calibrate event data provided through our tags. Identity of individuals or specific profiles are not necessary to provide robust planning tools or cross-platform audience estimates. Cookies, never a measure of persons (or a great measure of devices for that matter), provided some context, when viewed through the lens of panel data, to estimate the number of persons per segment, demographic and/or site. The enforcement of opt-in permissions related to ePrivacy and GDPR and the release of Apple’s Intelligent Tracking Protocol (ITP) have already provided benefits in limiting the exchange of user information without consent as the prevalence and scope of cookies for some segments have declined. For methodologies that depend exclusively on cookie identity and details, this change affected data quality and introduced bias in measurement, particularly with mobile. The Comscore panel-calibrated methodologies have been able to adjust to this change in cookies and still deliver quality person-based audience measurement.

The Privacy Sandbox and data exchange approaches still need validation and bias correction. The aggregate feeds can be calibrated based on Comscore’s panel so that they are accurate for planning purposes and validated for ad measurement. The secure exchange proposed by the community will help with the infrastructure to perform data validation, but Comscore’s panel and experience processing varied datasets are unique in being able to provide the capabilities the industry expects. Comscore has developed processes for integrating these server-side exchanges and validations. Some clients have taken advantage of this approach with the feeds coming in aggregate but using respondent-level panel data as a source of truth to validate the individual segments. These new aggregate feeds do not adversely affect the quality of data Comscore reports and will continue to grow in popularity as instrumentation evolves.

Ad Targeting in a Privacy-Focused Era
Ad targeting is a use case under consideration for the web standards community. Even with aggregate segments defined for a target to protect privacy, identification of the segment and an exchange of information still must be executed at some ID level. There are multiple companies that have come up with new identification spaces that do not rely on cookies but rather link information across publisher freewalls and paywalls. This will shift data exchanges into more explicit first-party, direct consumer consent arrangements between publisher ID spaces. However, these ID spaces are not complete and, until more universal adoption, offer another degree of fragmentation and potential misalignment. Aligning these ID spaces to provide consistent measurement and understanding of all audiences across all platforms is not any different from the process to calibrate individual sources. Again, Comscore’s methodology and panel assets are uniquely suited to join across and normalize these spaces for a true, person-level, holistic view. This not only helps to ensure consistent audience measurement and validation but can also help bridge the gap to ensure more consistent delivery in an activation setting.

Joining, calibrating, and validating across data assets and ID spaces addresses the mechanical shifts from third-party cookie frameworks to the new processes laid out in initiatives like the Privacy Sandbox. However, these initiatives do not address the need for robust, cross-media data assets to plan, segment and measure audiences in digital browser, video, and linear TV spaces. Even a panel as large as Comscore’s or the use of universal ID assets will not give the breadth and depth of coverage required by the industry. To answer these questions, a new way of measuring audiences and thinking about segments is required.

Looking to the Future
Comscore is at the forefront of the research to apply inference techniques on top of deterministic data assets in order to complete measurement. Over the next year, emerging research to take real events but combine them into aggregate representative respondent profiles will yield a new audience measurement that we refer to as our “Atomic ID.” This Atomic ID becomes the base factor for defining audiences and behavior. It provides the granularity to understand cross-domain and cross-platform behavior and profiles while remaining aggregate enough to not tie to any single individual or user. The IDs can make use of panel calibrations to integrate various publisher-level datasets, as Comscore has done for the last 10 years. With these IDs applied at Comscore’s census scale, next generation granular cross-platform reporting in MMX® and vCE® will transition to “empaneled” IDs to be a representative audience asset, the Census-Based Panel. Looking even further down the line, these IDs can be combined with observed propensities from panels or other assets and become an infrastructure for what-if analysis and simulation sandboxes, providing a deeper view into planning, lift, and measuring advertising ROI.

The Google announcement starts the clock on a shift in infrastructure from convenient, non-controlled cookie-based identity management and exchange to a more structured environment where consumers have more direct control of their data and metrics are more well-defined. This new environment will come with significant advantages and fix problems that exist today with ad targeting, validation, and audience measurement, but it will not fix all challenges and may even create some new ones. In the end, we will continue to need to understand how these measurement feeds relate to real person behavior and how we can calibrate to the true digital population. Comscore has the experience, data assets, and integration capabilities to continue to provide that measurement to the digital ecosystem.

To learn more about Comscore measurement in a cookieless world, click here.