PurposeDigital intermediaries such as Google and Facebook are seen as the new powerbrokers in online news, controlling access to consumers and with the potential even to suppress and target messages to individuals. Academics, publishers and policymakers have raised concerns about the implications of this new power, from the impact on media plurality to implications for democratic discourse, freedom of speech and control over public opinion formation. After reviewing academic literature that has raised this concern and public policy addressing it, this paper aims to examine the empirical foundations for these claims. Through secondary analysis of industry data on referrals of online news traffic, the authors find that intermediaries do have the potential to exert significant influence over distribution of online news. The authors however find that not all news that is filtered through intermediary services is subject to the same shaping and editorial forces, in part, because user agency is also an important factor. The role of intermediaries in news distribution is thus complex; headline numbers do not translate automatically into influence due to the complex interplay and exchange between user agency and the editorial influence of intermediaries.Design/methodology/approachThis paper is based mainly on a secondary analysis of publicly available data on news referrals, and some data provided by news publishers along with re-analysis of regulatory data from Ofcom and previously unpublished data from the BBC and SimilarWeb. These data sources are combined for the first time to investigate claims regarding the current controversy about media plurality, algorithmic power and transparency.FindingsThe paper finds that evidence that intermediaries wield concentrated editorial power is mixed. While other, non-intermediated news distribution platforms such as TV and the press remain highly important, online is heading towards being the most important distribution platform, particularly for younger demographics. The authors found that intermediaries such as search and social control access to a significant proportion of online news content. Not all use of intermediaries is indicative of online gatekeeping however. User agency also determines how content is prioritised and thus consumed. The news consumed is therefore a product of a complex interplay between user agency and intermediary influence. In contrast to traditional discussions of media power and its regulation (for example the notion of mass media plurality); it is thus not possible to make inferences on influence simply by noting the market share of intermediaries. The role of intermediaries is much more subtle and opaque.Research limitations/implicationsThis paper is mainly based on publicly available data. It is crucial to find out what is possible with such data as regulators with responsibility for monitoring and regulating media plurality are similarly limited to such data. The implications are that further research with a wide range of methods and data sources will be necessary to update research on media plurality and diversity.Practical implicationsThe implications of these findings are that independent public authorities should have access to much more revealing data about public opinion formation processes, including referrals and other data currently held only by publishers. The three stage analytical framework will be of use to regulators and policymakers currently looking into these issues.Social implicationsCivil society and public debate about digital intermediaries is currently intensively discussed in policy debate. Taking these debates forward will depend on whether existing public policy frameworks (such as limits on news plurality) are able to accommodate the new challenges such as intermediary influence on news distribution and public opinion formation.Originality/valueThe recent special issue ofINFO, including contributions from Mansell and Helberger, raised a range of similar issues with regard to media plurality and intermediaries. These papers did not seek empirically to examine in depth, using all available publicly relevant data, the implications for media pluralism and diversity in one particular media market. The paper is theoretically original, contains some previously unpublished data and an entirely new empirical and theoretical analysis. The models and tables are previously unpublished.