Brave has pioneered a new breed of privacy-respecting targeting that is matched directly on the user’s device (aka “client-side”), without any personal data phoning home to Brave’s servers. Yet Brave Ads also offers standard reporting on ad impressions and engagement, and still allows for many standard measurement practices so advertisers can properly measure ROI.
Brave uses a range of cryptographic technologies and on-device machine learning to ensure that no personally-identifiable data ever leaves the user’s device. Contrary to other major ad tech providers who suck up as much personal data as they can, Brave only ever sees anonymous data.
Each of Brave’s four ad units has a unique approach to achieve varying levels of ad relevance with the end user. Here’s a summary how each ad unit works:
Search ads, unlike other Brave Ads formats, are not browser-based. They use country, device type, and—most importantly—keyword matching to privately pair users with relevant search ads.
Notification ads target users based on their browsing and search history, using Brave’s most sophisticated ad targeting system. Notification ads are matched entirely on a user’s device with a machine learning model that identifies contextual and behavioral signals and pairs them with ads. The end result is that data never leaves the device, meaning there isn’t a compromise between user privacy and ad relevance.
The matching engine works using a distributed catalog system. Approximately every hour, a catalog of ads and their targeting parameters are sent out to all devices. Different countries and regions receive different catalogs—which reduces the load of ads delivered to each device. Each ad has a profile of targeting parameters, which allows it to be matched anonymously with user profiles. This allows users to receive relevant ads without risking their personal data.
Newsfeed ads are targeted based on country, language, and device type.
New tab takeovers
New tab takeovers are targeted based on country.