Consent and Choice: Experimental Evidence on User Behavior at Consent Banners

Cookie banners appear on websites when visitors first arrive, informing them about a site’s use of cookies and tracking technologies, and often asking them whether they consent to their data being collected and shared with third parties. We use a panel of 789 U.S. Chrome users, whose online behaviors were tracked by a browser extension they kept installed for up to three months, to investigate how users interact with cookie banners in the wild, over time, across all the websites they visit. We find that users engage with only one in three banners they encounter, and do so mostly on websites that force a response; these websites register three times more banner engagement than the ones where banners can be ignored, but lose more than twice as many visitors after their encounter with the banner. Users’ explicit choices (accepting or rejecting tracking) exhibit across-user heterogeneity, and are significantly influenced by whether a one-click reject option is offered. We set a single signal from the choices users make during the first third of the days we observe them, and find that this signal is consistent with about two thirds of their later encounters. We calculate that users would spend 58 percent less time on banners if websites could not force a response.