Monitoring of public and political discourse related to the elections on Facebook during the week leading up to Election Day
Published Sunday 19 April 2026 at 23:47
On 19 April 2026 Bulgaria held parliamentary elections against a backdrop of persistent political instability and documented disinformation activity.
This report presents the results of a structured monitoring exercise covering Facebook activity related to the Bulgarian parliamentary elections of 19 April 2026 in the week leading up to and including election day. The monitoring period runs from 12 to 19 April 2026 and focuses on posts matching the search string изборите ("the elections"), retrieved via Meta's Content Library research platform. The project analyses the structural composition of that content — volume, producers, content types, reach, and engagement — and applies a purpose-built keyword detection framework to identify posts carrying recognisable narrative frames associated with disinformation and political influence operations.
The project is built on two distinct data sources produced by Meta's Content Library platform – the research interface of the Transparency Tools, filtered on the search string "изборите" for the period 12–19 April 2026.
The first one – a downloadable row-level research dataset – contains full post-level structured data — post text, creation timestamp (UTC), content type, surface type (page/group/profile/event), reactions, shares, comments, and views — but is subject to Meta's follower threshold rule: only pages with ≥15,000 followers and verified or sufficiently large profiles with ≥25,000 followers enter this subset.
The second one – а lightweight small summary/preview export from the same Content Library query – is a manual UI copy-paste export of the on-screen Content Library search results, which aggregates all matching posts regardless of producer size, but provides only truncated post text (~42–50 characters), a single aggregated reactions figure, views, content type, and producer name at day-level date precision.
The comparison between the two demonstrates that the Meta/Facebook follower threshold imposes a structurally asymmetric cost on small-language research contexts. A Bulgarian-language page with 14,000 followers represents a meaningful share of the national online public. An individual profile with 24,000 followers in Bulgaria operates at a scale that, in absolute terms, positions it among the country's more widely-followed public voices.
The 291 absent Bulgarian-language posts at the ≥10K views threshold generated by 210 unique producers accumulated 12.72 million combined views which remains out of sight when the removable subset restriction is applied. Several of those posts reached 200K–800K views from sub-threshold accounts, audience levels comparable to established media pages.
The structural finding that this stratum is dominated by individual profiles posting Status content (19% of the absent corpus versus 6% in the downloadable subset) reveals that the threshold systematically suppresses the individual civic voice layer of Bulgarian political discourse precisely at the moment of an election.
Read the full report here.
