Experts examined one million sites, creating a special program to overcome obstacles that require human interaction when working on the Internet. It finds a search string, uses it to search for JELLYBEANS, and then collects all subsequent network traffic for analysis. The idea was to examine each HTTP request and see if “JELLYBEANS” appears in requests forwarded to third-party partners.

what was discovered

• Understood that 81% of websites outsource customer searches to third parties.
• These sites cover every category imaginable – adult, shopping, travel, and even health.
• Search terms collected by these sources may include sexual preferences and gender identity, shopping habits, and medical information.

Norton Labs believes the true percentage of search query leaks is likely even higher, as the program written by the researchers doesn’t always “break” sites’ protection. That’s why they call the number 81 the “lower bound”, but actually the data consumes more resources.