How Pikasa's Election Scanner tracked the road to the 25 October elections through the resignation that upended the race

 

Overview

Between 1 and 27 September 2026, Analytics.Live tracked 48,066 articles and social posts about twenty leading Serbian political figures, generating 6.5 million engagements, 207 million video/post views and a cumulative reach of 772.9 million. Traditional media did the heavy lifting on volume and reach; social platforms did the heavy lifting on engagement.

Fig. 1  Media vs. social media, September 2026 · Analytics.live

News articles made up 82.6% of everything tracked (39,689 items) and, at 583.2 million, accounted for three-quarters of total reach a reminder that in Serbia's pre-election cycle, traditional outlets still set the pace. For scale, 53,724 articles ran under the general "Politics" category across Serbian media in the same period, so the figures we tracked by name alone accounted for close to three-quarters of all political coverage; almost any politics story published was likely to mention one of them. But social platforms, though only a fifth of volume, produced 5.1 million of the 6.5 million total engagements about 610 reactions, comments and shares per post, versus 35 per article. The two channels are not competing for the same job: articles carry the story further; social posts are where people argue about it.

The Context

Serbia entered the autumn of 2026 in the middle of its most consequential electoral cycle in years. Nearly two years of student-led protests triggered by the fatal collapse of a railway station canopy in Novi Sad in November 2024 had kept questions of accountability, institutional trust and media fairness at the centre of public life. As the ruling Serbian Progressive Party (SNS) and a fragmented opposition, including the student-backed "Students Win" list, headed toward snap parliamentary elections called for 25 October 2026, the information space around the vote was dense, fast-moving and, as our data shows, heavily uneven.

Then, on 26–27 September, the race changed shape entirely: President Aleksandar Vučić, constitutionally barred from a third presidential term, resigned the presidency to run for prime minister instead a move that handed the caretaker presidency to his close ally, Assembly Speaker Ana Brnabić, and turned the final month of the campaign into a referendum on continuity.

How Pikasa Reads Beyond the Headlines

Pikasa's political scanner was already tracking this landscape when the resignation broke, giving us a real-time, before-and-after view of how a single political event reshapes coverage, engagement and narrative framing.

Reading the full text behind the headlines not just the politican mentions adds texture the counts alone can't. A Faktor plus survey published twice in September put SNS at 48.6% of the vote if the election were held that week, with only two other lists projected to clear the parliamentary threshold: the Socialist Party of Serbia (SPS, 4.9%) and the "European Serbia" coalition (3.6%). Right-wing parties polled below the threshold, and the student-backed list registered only a modest month-on-month gain. A separate data point tracked in the same window from the prediction-market platform Polymarket showed Vučić as the heavy favourite to become prime minister after 25 October, well ahead of Democratic Party leader Dragan Đilas. Both signals point the same way: a fragmented opposition field facing a dominant incumbent machine.

Three Things the Data Shows

1. A saturated media environment with a narrow center of gravity

Serbia's pre-election coverage was dominated by a small cluster of high-volume outlets. Just five sources Informer, Danas, Politika, Alo and Novosti accounted for over a quarter of all tracked items, alone publishing 3,985 pieces in the period, more than any other outlet. Daily volume rarely dropped below 1,200 items and peaked at 2,188 on 4 September, underscoring how much of the campaign was being fought in earned and owned media simultaneously rather than through paid advertising alone. On 4 September the spike (2,188 items, the highest single day that month) was driven mainly by one story that dozens of outlets ran almost simultaneously: a new Faktor plus poll (fieldwork August–September, ~1,200 respondents) putting SNS/Vučić at 47.2%, the student-backed "Studentska lista" at 31.5%, SPS at 4.9%, and two other lists just clearing the threshold - SSP-led pro-EU coalition at 3.6% and the right-wing Narodni pokret Srbije at 3.4%. Dozens of near-identical headlines ran the same numbers within hours, which is what produced the volume bump rather than any single big event.

2. One name commands nearly half the conversation

Across every figure Pikasa Analytics tracked, President Aleksandar Vučić appeared in 43.2% of all coded items more than the next nine most-mentioned politicians combined. Prime Minister Ivica Dačić and Finance Minister Siniša Mali were a distant second and third. This concentration of coverage is itself a finding for anyone assessing the fairness of the pre-election information space: visibility in Serbian media before 25 October was not evenly distributed across the ballot.

3. The resignation didn't spike volume it spiked attention

Counter-intuitively (even for a weekend), the two days around the resignation (26–27 September) produced fewer articles and posts than the monthly average 1,242 items on the 26th, the lowest single-day count of the month, rising to 1,684 the next day. But engagement told a different story: the 26th generated 195 engagements per item, against a monthly average of 135, and drove 10.2 million views, well above the daily norm. Fewer stories were published, but far more people engaged with each one, because raw volume alone would have missed it entirely.

The framing shift was just as telling as the numbers. Some outlets reported the resignation as a constitutional and political event; other pivoted within hours to campaign messaging built on the same news:

That same-day pivot from resignation announcement to campaign narrative, on the same masthead is exactly the kind of coordinated framing shift that manual monitoring struggles to catch at scale, and that Pikasa's methodology is designed to surface.

Why It Matters

For newsrooms, election observers, civil society organizations and donors working in Serbia and the wider Western Balkans, the month before 25 October is a case study in how quickly an information environment can turn on a single event and how misleading raw headline counts can be on their own, without engagement and framing analysis layered on top.

This is the gap Pikasa’s Election scanner is built to close: combining Informer, political scanner, topic scanner into a single view of how political narratives move, who drives them, and when a routine news cycle turns into something that needs closer attention.

Methodology

Using Analytics.Live automatically were tracked every article and social post mentioning twenty of Serbia's leading political figures across seven channels news articles, Facebook, Instagram, YouTube, TikTok, Telegram and LinkedIn from 1 to 27 September 2026. In total, the monitor captured 48,066 items from over 100 distinct outlets and accounts, generating 6.5 million engagements, 207 million views and a cumulative reach of 772.9 million across the period. For comparison in this period, Pikasa Analytics monitored more than 230 digital media outlets who published more than 184k articles, and more than 8k social media public channels who produced more than 350k posts.