Bottom line
Cloudflare, the company that runs a large chunk of the internet's traffic-routing infrastructure, looked at its own network data during the 2026 World Cup and found a clear pattern: local kickoff time, not the stage of the tournament, was the biggest driver of internet traffic swings. Night matches sent traffic surging as people stayed up to watch. Daytime matches barely moved the needle, or even pushed traffic down, because people were already online for other reasons. The single match that shook the internet hardest, by Cloudflare's own measure, was a quarterfinal, not the final. This is a company blog post built on proprietary traffic logs, not a peer-reviewed study, so treat it as a solid, data-rich observation rather than a scientific finding with confidence intervals.
What they looked at and how
Cloudflare operates points of presence in more than 330 locations across over 120 countries, giving it visibility into a large share of global web traffic. For this analysis, published on Cloudflare's blog on July 21, 2026, and picked up by IT-Connect on July 23, the company built a per-minute baseline from the median traffic of the previous four weeks, for each country. Actual traffic during the tournament was then compared to that baseline and expressed as a log₂ ratio: zero means normal traffic, +1 means traffic roughly doubled, −1 means it roughly halved. There is no independent peer review here and no stated funding or conflict beyond Cloudflare analyzing its own commercial network, which is worth keeping in mind since the exercise also serves as a showcase of Cloudflare's visibility into internet-wide behavior, according to Cloudflare's account.
What they found
The strongest pattern was time of day. Matches kicking off roughly between midnight and 8 a.m. local time produced the largest traffic spikes, sometimes more than doubling normal traffic, according to Cloudflare. Daytime matches, by contrast, barely shifted traffic, since most viewers would have been online anyway.
Bosnia and Herzegovina illustrates this well, per the same analysis: when the national team played at 2 a.m. local time, traffic rose well above normal, even doubling. When it played in the evening, traffic instead dropped to about 70% of its usual level, as people set screens aside to watch the match directly.
The Brazil-Japan round of 16 match on June 29, 2026 (Brazil won 2-1) shows the same effect across two time zones twelve hours apart: in Japan, where the match fell in the middle of the night, traffic rose to roughly +1 on the log₂ scale, about double normal. In Brazil, where the match fell during the active part of the day, traffic dropped to about −0.4, meaning it fell to roughly three-quarters of its usual level, a drop of about 24%.
To rank individual matches, Cloudflare measured the absolute deviation from normal traffic in a two-hour window after kickoff, for each country, excluding matches played at the same time as another (since a spike could not be attributed to one game or the other). By this measure, the match that moved global traffic the most was not the final or a semifinal. It was the Argentina-Switzerland quarterfinal on July 11, won 3-1 by Argentina, with a deviation factor of about 1.26, ahead of the France-Spain semifinal at 1.21. The rest of the top matches were a mix of quarterfinals, round of 16, and even round of 32 games. At the team level, Argentina came out on top too, with a factor of 1.17: when Argentina played, traffic in a typical country deviated by roughly 17% from normal, the largest pull of any team in the tournament.
What this doesn't show
Cloudflare's exact mathematical definition of the "factor" used for the match and team rankings (1.26, 1.21, 1.17) isn't spelled out in the material available here {{TODO: verify}}, so it's unclear whether it's the same log₂ scale used elsewhere or a separate metric. The analysis covers HTTP traffic passing through Cloudflare's network, which is large but not literally all internet traffic; video streaming or traffic routed through other providers may behave differently and isn't captured. The detailed country examples given are limited to Bosnia and Herzegovina and the Brazil-Japan pairing, so it isn't clear how uniform the pattern is elsewhere. Nothing here separates the effect of kickoff time from other factors that might coincide with it, such as public holidays or how far a national team advanced. And as a company blog post rather than an academic paper, there is no statistical testing of whether these deviations exceed normal week-to-week noise.
What it means, if anything
The consistent piece of information here is straightforward: what pulls people offline, or keeps them up, during a global event is mostly when the match falls in the local clock, not how big the match is on paper. The Argentina-Switzerland result is a useful reminder that "important" and "disruptive" aren't the same thing when it comes to collective online behavior. Beyond that, this is a descriptive look at one company's traffic logs during one tournament, interesting on its own terms but not a basis for broader claims about internet infrastructure or audience behavior generally.
