A compiled global digital report is not one platform's disclosure

The test

DataReportal, produced by the analytics firm Kepios, published its Digital 2025: Global Overview Report on 5 February 2025, in partnership with We Are Social and Meltwater. The report states it draws on data supplied by named partners including GWI, Statista, GSMA Intelligence, Similarweb, Semrush, data.ai and Ookla, alongside United Nations population data, and that its analysis was written by Simon Kemp without generative-AI drafting.

What the evidence says

The report itself demonstrates why sourcing matters: it states that Kepios's analysis puts 'global social media user identities' at 5.24 billion, explicitly noting that user identities may not represent unique individuals, since one person can hold accounts across several platforms. Separately, the report walks through a live discrepancy in TikTok's numbers: ByteDance's own ad-planning tool showed potential reach falling by almost 100 million accounts between October 2024 and January 2025, a 5.7 percent quarterly decline, while independent usage data the report cites from GWI, data.ai, Similarweb and Semrush showed no corresponding drop in active use. DataReportal's own editorial judgment, stated in the report, is that the ad-reach decline likely reflects ByteDance removing duplicate or inauthentic accounts from its own reporting tool rather than a fall in real usage, a distinction the report draws explicitly between a platform's self-reported ad-audience metric and third-party usage estimates.

The sample and the variance

Because the report is a compilation, each figure in it carries the measurement method of whichever partner supplied it: mobile app data from data.ai, survey data from GWI, search and traffic data from Similarweb, and platforms' own advertising tools for ad-reach numbers. These methods disagree by design, ad-reach tools count potential advertising audience, not confirmed active users, while panel and app-tracking data measure different populations again, so a single 'users' figure attributed only to 'DataReportal' obscures which method actually produced it.

What to try next

A marketer or researcher citing a DataReportal figure should name the underlying data partner and method the report attributes to that specific number, rather than citing 'DataReportal' as if it were a single measurement instrument. This is a sourcing discipline the report's own text repeatedly models, not an additional editorial layer imposed on it.

• Which named data partner, or which platform's own ad-planning tool, does the report attribute this specific figure to?
• Is the number a self-reported platform ad-reach figure, a third-party usage estimate, or a Kepios-modelled total?
• Does the report's own text flag a discrepancy or caveat around this figure, as it does for TikTok's 2025 ad-reach numbers?

DataReportal's value lies less in any single headline statistic than in its practice of naming its ten-plus data partners and, at points, showing where they disagree, a habit worth preserving whenever one of its numbers is lifted out of context.