Auction model aims to improve consumer data quality

Businesses increasingly depend on consumer data for strategy, market analysis, product development and AI recommendations, but widely available datasets are often biased or unreliable. Privacy tools such as ad blockers, VPNs and incognito browsers make behaviour harder to track, while rules including GDPR and CCPA reinforce consumer control over personal information. Data collected from people who are easier or more willing to track can overrepresent less privacy-sensitive consumers, weakening market insight and leading to mispriced services, mistargeted campaigns and products customers do not want.

Consent-based data markets can help, but common approaches have trade-offs. Fixed compensation is cheaper but tends to miss privacy-sensitive users, while the centralised optimisation model can prompt consumers to exaggerate the price they require to share data, pushing companies to overpay.

A proposed mechanism called Random Sampling of Rolling Pairs matches subjects who appear identical in their data but differ in privacy concerns. The platform compares private compensation demands through a Vickrey-style process, selects the lower bidder, and pays that person the higher losing bid. Simulations using real world data found the approach can collect unbiased, reliable data at near-optimal cost while supporting compliance with privacy rules. Stronger regulation of data sellers and resellers is also needed to make data markets fairer and more reliable for companies, AI systems and consumers.