OCTOBER 5, 2022 | MPC DATA PRIVACY SUMMIT
MPC & the Ads Industry
The ads ecosystem today is evolving due to an increased demand for privacy and more regulation. At Meta, we care deeply about enabling the best advertising experience, which means connecting people to products and services that they will love, and allowing businesses to measure the impact of their Meta ads. That’s why alongside academics, global organizations and developers we’re investing in a multi-year effort to build solutions using privacy-enhancing technologies, like multi-party computation. These technologies will help us minimize the amount of personal information we process, while still allowing advertisers to reach relevant audiences and measure ad effectiveness.
MPC & Meta Ads
For the last year, Meta has been testing MPC for ads measurement, with our Private Lift and Private Attribution solutions. Through the use of MPC, advertisers can understand how their campaigns are performing while limiting what the advertiser and Meta can learn about a person. Additionally, we’ve improved the back-end protocol for MPC – which makes it easier for other businesses to build MPC into their privacy solutions – driving better performance using less network traffic. All of these things greatly reduce the cost of implementation, which is why we open sourced our Private Computation Framework code, so that other businesses can adopt this state-of-the-art protocol and algorithms to more efficiently use MPC.
MPC & Ads Industry Standards
These challenges are not specific to Meta and impact the entire ads industry. This is why industry groups like the World Federation of Advertisers (WFA) and the World Wide Web Consortium (W3C) are working to co-build solutions and standards to address these challenges. For example, when advertisers buys ads, they want to know how many times those ads led to a purchase of one of their products. Historically, this has been done by linking cross-site user data with global identifiers like 3rd party cookies. As browsers make changes to prevent tracking, we are working together as an industry to develop new, more private approaches. Interoperable, Private Attribution (IPA) is one proposal for how browsers could enable attribution in MPC on secret shared data. (Read our proposal here.)
- Interoperable Private Attribution
- Private Matching for Compute
- Privacy-Preserving Randomized Controlled Trials: A Protocol for Industry Scale Deployment
- The Privacy-preserving Padding Problem: Non-negative Mechanisms for Conservative Answers with Differential Privacy
- Multi-key Private Matching for Compute
- Private Computation Framework 2.0 whitepaper
Open Source Libraries