What's the 800-Pound Gorilla in the Privacy Landscape?

How to maximize the single-source panel

Panels should be shepherding in this new era of data transparency—while simultaneously providing the necessary layer of person-level metrics that will soon become table stakes across the industry. We’re starting to get a good look at how high and precarious those stakes are.

On Sunday, Rep. David Cicilline reportedly said he wants to “hit Big Tech with the legislative equivalent of a swarm of drones” this year. With the Virginia Consumer Privacy Data Protection Act just signed into law (mirroring the late 2020 passing of the California Privacy Rights and Enforcement Act), it’s a matter of time before more state regulatory commissions circle the wagons and accurate data will take on an even greater value. Florida is already in the wings.

Given this critical juncture, the privacy-safe, future-proof ace kicker has been under our noses for decades: panels. Everybody uses them. Panels are the 800-pound gorilla in the room the industry is afraid to talk about because they don’t have the curb appeal of set-top box and smart TV data. But curb appeal doesn’t stand up to close scrutiny in the age of tightening privacy and looming antitrust legislation.

Panels represent that bridge between device-delivery data (i.e. census) and person-level data that is fully permissioned for use in the cross-media/advertising ecosystem. Recently, the World Federation of Advertisers (WFA) proposed an industry framework for cross-media measurement that would serve as a “North Star” for advertisers. Panels are the first of four key components in the proposal that provides a single source of data that includes granular individual-level data that has been collected with “explicit verifiable consent.”

In other words, single-source panels act (as the WFA says) “as the arbiter of truth, providing benchmarks for the use and overlap of media consumption across channels and screens.” To maximize the value of panel data to calibrate device delivery data, it must definitively measure all persons, TVs and digital devices within a panel home.

In the end, it would not be nearly as powerful or valuable as true persons-level from a single-source panel. It would be like asking world champion driver Lewis Hamilton to race his Mercedes-AMG F1 with low-octane fuel; it simply would not get us to the finish line of “a census identification framework and a private reach and frequency estimator.”

Given this new paradigm, these are three foundations of panel recruitment that media executives should know if they want to get the dead-on truth from the viewing public.


When recruiting participants into a panel, every recruit is fully informed about the data being monitored, including their TV, smartphones and laptops. Often, prospective panelists are surprised that this information is not already “known” about them (“I just figured [big data company] already knew this”). This “upfront” approach and letting them know they have the right to withdraw their permission at any time is a true value proposition for building a panel. 


From the start, I tell potential panelists about research participation, including how I will do it and where the data will go: “We’ll always aggregate your data. You are not going to get advertisements directed to you for participating.” This level of transparency builds the trust that allows for long-term participation. 


In measuring media, panel incentives are strictly behavioral. Incentives do not increase because you’re spending more time with media, but based on joining the panel, length of participation, level of cooperation, and the amount of things being measured to deliver good data. We understand that there is a trade-off for agreeing to share your data. So if privacy concerns are met, participants receive a direct benefit.

Media stakeholders will need to be more careful than ever with their data. It has never been more important to understand how data was collected, what it truly represents and how it fits into their larger data strategy.