The Fine Print

Method

The dataset covers 22 very large online platforms for the second half of 2025, the first cycle reported under Commission Implementing Regulation 2024/2835. We retrieved each provider's published filing from its source, loaded it into a Postgres database as filed with one figure per source cell, and harmonised it through numbered SQL migrations. We normalised the form of the data and recorded each change and kept the original. The transforms are listed in Footnotes. The derived comparison views over the harmonised data are served as a read-only public API, which this site reads at build time.

The designated search engines are not yet included. Banded audience figures and restatement versions are reported as filed and noted in Footnotes.

Three providers filed a document beyond the eleven harmonised files. AliExpress and TikTok published narrative reports on their processes and methods, and Google reported its recipient figures in a separate document covering all its services. None of these form part of the comparable dataset, and each is noted on the provider's page.

The loaders and numbered migrations are in thedata repository.