Brazil’s Runoff Election Left Clear Marks on National Traffic Patterns
Brazil’s 2022 presidential election drew worldwide attention, and not just for its political stakes. The two voting rounds—the first on October 2 and the runoff on October 30—produced measurable shifts in how Brazilians used the Internet. Cloudflare’s network data shows consistent dips in overall traffic during voting hours, strong interest spikes around result announcements, and notable differences in behavior between the two rounds.
All times below are local (UTC-3), reflecting Brazil’s most populated regions.
Traffic Dips While Polls Are Open
As seen in other countries with Sunday elections, Brazilian Internet traffic dropped while voters headed to the polls. The dip was more pronounced in the runoff: traffic fell 21% on October 30 around 18:00, versus a 10% drop on October 2 near 20:00. Turnout was also slightly higher in the second round, with 124 million votes cast compared to 123 million in the first.
Comparing all five Sundays in October makes the pattern clear. Election Sundays stand apart from the other three weekends, with identical shapes despite different magnitudes.
The effect appeared across major Brazilian network providers. Both Claro (AS28573) and Telefonica (AS27699) showed the same midday slump on October 30, as did traffic from key states including São Paulo, Rio de Janeiro, Minas Gerais, and Bahia.
Mobile Share Rises on Election Days
Mobile devices made up a larger share of Brazilian traffic on voting weekends. The three highest mobile percentages in October were October 2 (66.3%), October 9 (66.4%), and October 30 (65%). This matches trends observed in previous elections, where users gravitate toward phones rather than desktop connections.
Election Sites Saw Stronger Pull in the Runoff
DNS query data from Cloudflare’s 1.1.1.1 resolver, measured against a September baseline, shows how interest in election-related domains evolved. The first round generated a 16x spike at 20:00 on October 2. The runoff peaked higher and earlier—at 17.4x above baseline by 18:00 on October 30, as results began arriving faster than in the first round.
Initial activity on October 30 started around 17:00 as polling stations closed, with a rapid climb to the peak within the hour. Results flowed more quickly in the runoff: Lula first took the lead at 18:45 with 67% of stations reporting, and was declared the winner near 20:00.
Candidate Domains: Divergent Patterns
Traffic to Lula-related domains clustered around key moments. A 13x spike hit on October 1 at 21:00, followed by two notable jumps on October 2—16.7x at 09:00 and 10.7x at 21:00 as the runoff became a certainty. On October 30, a single dominant spike at 16.7x above baseline coincided with the winner announcement around 20:00.
Bolsonaro-related domains showed a different rhythm. Activity was elevated in the days before the first round, hitting 10x on September 30, then spiking at 8x during results on October 2. But the month’s largest surges came later: 20x on October 16 after the first televised debate, and 22x on October 25 during the final campaign week when polling predictions were released. The second debate on October 28 produced smaller bumps for both campaigns.
News Sites: First Round Drew More Interest
News organizations saw heavier traffic during the first round than the runoff. DNS queries peaked at 9.1x above the September baseline on October 2, versus 6.1x on October 30. Timing differed too: results arrived earlier in the runoff, pushing the news traffic peak to around 19:00 rather than 20:00.
By 19:45 on October 30, with over 98% of votes counted and Lula’s victory clear, news traffic began to drop sharply. The first round dragged on longer—it wasn’t confirmed until around 22:00 that a runoff would occur—and news traffic stayed elevated at 8x through 21:00 and 22:00 before tapering.
Real-World Events Shape the Internet
The election data reinforces a familiar pattern: collective events, whether political, sporting, or cultural, leave identifiable signatures on national Internet behavior. Beyond the obvious dips and spikes, the timing and intensity of those changes offer insight into how populations engage with unfolding events in real time. Traffic shifts don’t just reflect what happened—they chart when people paid attention.
