A common format changes the comparison
The first benefit of harmonised reports under the Digital Services Act (DSA) is methodological: common categories and formats reduce some of the difficulty of comparing documents that previously organised information in different ways. The European Commission said the publication deadline for the first round of harmonised reports was the end of February 2026. The Commission says the new format makes it easier to compare moderation data across services. That improves comparability, but it does not mean that every figure describes exactly the same operational process, or that every difference between platforms has an obvious explanation. European Commission
The legal basis for the templates is Implementing Regulation (EU) 2024/2835, which establishes models for the reporting obligations of providers of intermediary services and online platforms. The Commission set 1 July 2025 as the date on which the new harmonised rules would apply. The change concerns the reporting framework: the regulation defines formats and information requirements, but that is not equivalent to independent certification of every figure a company reports. It is therefore important to distinguish the uniformity of the template from verification of the data entered into it. Implementing Regulation (EU) 2024/2835 European Commission: harmonised rules take effect
The timetable also matters when interpreting the results. According to the Commission, harmonised reporting periods are six months: 1 January to 30 June and 1 July to 31 December, with publication expected at the end of August and February, respectively. To compare two figures, readers need to confirm that they cover equivalent periods and take account of when the format began to apply. A change between periods may prompt a question, but by itself it does not show that moderation intensity has increased or decreased: activity volumes, reported categories, or the way decisions are recorded may also have changed. European Commission: harmonised rules take effect
What can be compared—and what should be kept separate
The reports bring together information about moderation practices using common fields defined according to provider type. The Commission has explained that reporting obligations may cover, among other things, the number of items of content removed, the accuracy of automated moderation systems, and account suspensions. Some reports also contain data about notices received and orders issued by authorities. The value lies in being able to ask comparative questions on a more consistent basis than documents prepared using disparate structures would allow. Before comparing two figures, however, readers need to check which field they are looking at and how the report defines it. European Commission: reporting rules European Commission: first deadline for certain platforms
A total is not automatically a violation rate, the prevalence of a problem, or the amount of harmful content circulating on a service. Interpreting a volume requires knowing the unit counted, the period covered, the categories applied, and the type of decision included. A notice received is not the same as content removed; a removal initiated by a platform is not interchangeable with a measure taken following an authority’s order. Data about automated systems describe one specific dimension and do not necessarily cover every decision involved in a case. Responsible comparison therefore preserves each datum’s definitions and context rather than reducing them to a simple platform ranking.
The scale of a service also matters. A platform with more activity may record more notices or decisions, but that figure alone does not show that it moderates better or worse than another service. Comparing performance requires suitable denominators and context, as well as an understanding of classification criteria. A common format makes it more feasible to study certain data side by side; it does not make every count an equivalent indicator. This distinction matters both when comparing companies and when examining one platform’s figures over time.
A second source: the database of statements of reasons
The reports are not the DSA’s only transparency tool. The Commission’s transparency database allows users to search statements of reasons submitted by online platform providers about moderation decisions. Its interface offers searches for individual statements and filters; the search page explains that available results correspond to data submitted during the previous six months, in line with the data-retention policy. That window needs to be kept in mind when comparing database records with a report covering a different period. DSA Transparency Database: search statements
The relationship between the two tools can help users frame more precise analyses. The Commission says that, from 1 July 2025, the categories and keywords used in statements of reasons were harmonised with the categories in the reports. The aim of that alignment is to make comparisons and consistency checks easier. In practice, analysts can investigate whether certain aggregate patterns in a report appear to correspond to decision records published in the database, while bearing in mind that the sources differ in scope and unit of observation. Matching categories help bring the sources closer together, but do not make them identical. European Commission: how the DSA improves transparency
The database also offers a research API for programmatic queries, as well as public browsing tools. Its documentation describes access to and querying of statements of reasons for academic research and public-policy purposes. These functions make it easier to examine large datasets, but do not by themselves ensure that a query represents all of a platform’s moderation activity. A statement documents a reported decision; on its own, it does not prove that the decision was correct, applied consistently, or representative of all comparable cases. DSA database research API
How to read a discrepancy without overreacting
If an aggregate report and the database of statements appear to diverge, the difference may be a reason to investigate, not automatic proof of non-compliance or manipulation. Before drawing a conclusion, readers should check the periods, categories, definitions, coverage, and units counted. For example, a table may summarise a volume for a complete reporting period, while a database query is limited to records still available within its retention window. An apparent divergence may reflect those limits on comparison; identifying the cause requires examining the specific case.
The same caution applies to labels and methodological choices. An analysis should state which fields it uses, which period it selects, and how it handles incomplete or duplicate data when relevant. The API documentation, for example, warns that statement search results cover the previous six months of submitted data and that the database is updated daily. These operational details do not invalidate the tool, but they do define what can be inferred from a particular query. Harmonisation makes it easier to ask questions and identify discrepancies that warrant explanation; it does not replace an audit or prove by itself that an irregularity exists. DSA Transparency Database: search statements
Nor is it sound to infer that a platform reporting more decisions necessarily moderates more or less effectively than another. Volumes may depend on a service’s size, detection mechanisms, content type, and applied criteria. Without comparable denominators and sufficient context, ranking services by raw figures can be misleading. The strongest interpretation is limited: standardisation makes certain fields easier to compare, but results must be read in light of what those fields measure and what they leave out.
What the figures do not cover
Reports can provide data about procedures, measures, removals, notices, and automated systems, but they do not tell the full story of a decision. An aggregate count does not necessarily explain the context of each post, the quality of human review, the impact on the person affected, or whether an appeal changed the outcome. Assessing those aspects would require additional data and methods, not just a common template. The regulation establishes reporting models; the existence of those fields is not an assessment of the merits of every individual decision. Implementing Regulation (EU) 2024/2835
Transparency is not the same as independent verification either. The official sources consulted describe the framework, fields, and tools available; they do not certify that every company calculates every indicator identically in practice or assess the accuracy of every report. That is a limitation of the available evidence, not a conclusion that the data are wrong. For the public, the immediate value is being able to consult platforms’ disclosures more easily and observe patterns across more than one source. For researchers, common structures and access tools facilitate reproducible analysis, provided methodological choices are documented. For regulators, changes or discrepancies can help frame questions and guide oversight.
The firmest conclusion must remain within these limits: the DSA and its harmonised formats make certain moderation data more legible and comparable, while the database of statements of reasons offers another way to examine reported decisions. But the reports alone do not prove the overall quality, fairness, or effectiveness of a platform’s practices. Understanding them requires reading definitions, periods, and counting units, and distinguishing what is observed from what is inferred. Their main contribution is to improve the possibility of scrutiny; turning that possibility into a conclusive assessment requires further analysis.