RN
Every article of this kind rests on a single document, and almost none of them say so.
The country rankings, the "most searched term in Sweden" trivia, the year-on-year gender splits — they all originate in one operator's annual traffic report, published since 2013 and still running. It is genuinely interesting, and it is a marketing product that a company writes about its own visitors. Those two facts coexist perfectly well, and the second one is left out of essentially every article that quotes the first.
That matters more in 2026 than it did in 2019, because the thing that report measures has been structurally broken in the meantime.
The measurement problem, stated plainly
A national ranking of who watches the most requires knowing what a country's population does. What actually exists is a count of visits to one set of domains, from IP addresses that resolve to a country, by people who were not blocked and did not route around anything.
Three separate distortions sit inside that.
Market share is not behaviour. In any country where the dominant platform is a local one, the report registers a low figure — not because less is watched but because it is watched elsewhere. Every ranking of this type is partly a map of where one company happens to be popular.
The traffic is not the population. A comparison across countries with wildly different connectivity, device sharing and language coverage is comparing measurement conditions at least as much as appetite.
The proxy layer. VPNs put viewers in whichever country their exit node sits in. That has always added noise, and it is no longer noise.
What broke it
Following the US Supreme Court's June 2025 decision upholding Texas's age-verification statute, and the wave of comparable state laws behind it, the largest operator in the space chose to disable access across a large bloc of US states rather than implement identity checks. Reporting puts that bloc somewhere between nineteen and twenty-five states, covering more than a third of the American population.
The traffic did not evaporate. Around the January 2025 effective dates in Florida and South Carolina, searches for "VPN" more than doubled.
Follow that through to what it does to a dataset. A large share of American viewing has either moved to other platforms that the report does not see, or has been relabelled as coming from wherever a VPN exit node happens to be. Both effects push in the same direction and neither is measurable from outside. Any 2026 statement of the form "Americans watch X% less than they did" is describing a blocking decision, and any statement about the countries hosting popular VPN endpoints is describing displaced traffic that was never theirs.
The same dynamic applies to the UK after its own age-assurance rules took effect in July 2025, and to every jurisdiction that follows. The measurement instrument is now being deflected by the thing it is trying to measure.
The money figure that everybody prints
There is a valuation that appears in almost every article on this subject, and it is worth tracing because the trail is short.
It comes from a 2015 US business news piece, which quoted a sociologist offering it as an estimate. She said, in the same coverage, that the industry's financial information consists largely of estimates rather than firm numbers. Since then it has been reprinted often enough to acquire the authority of a measured figure, which it never had.
The range of published estimates runs from single-digit billions to the high tens of billions. A spread that wide is not a margin of error. It is an admission that nobody knows, and it is what you would expect from a sector combining private companies, offshore incorporation, cash businesses and a large free tier funded by advertising that nobody outside the ad networks can audit.
The honest version is that the industry is large and its size is not publicly established. Any specific number offered with confidence has been copied rather than calculated.
What can still be said
Discarding the numbers does not leave nothing. It leaves the structural observations, which are more durable than any ranking and were always the interesting part.
Prohibition redistributes rather than reduces. This is the most robust finding available, and the 2025 US blocks demonstrated it in real time — access restricted, VPN interest doubling within days, traffic reappearing elsewhere. The predictable effect of an access barrier is to move viewing toward platforms with less moderation and less incentive to comply with anything, which is the opposite of the stated intent.
Language shapes catalogues more than taste does. Where a country's search patterns look distinctive, the explanation is often that the available material in that language is distinctive, which is a supply fact wearing a demand costume.
Local platforms dominate where they exist. The countries that look like outliers in global data are frequently the ones with a strong domestic service, and the outlier is in the dataset rather than in the population.
Mobile changed the shape of consumption. Sessions got shorter and more frequent as viewing moved to phones. This is visible across the whole internet and is not specific to this category.
None of those require a percentage to be true, and none of them will be invalidated by next year's report.
What this page will not tell you
It will not rank your country, because the ranking would be one company's market share wearing a flag.
It will not list what your country searches for, because those lists are drawn from the same source and are now filtered through geoblocking and proxy displacement that nobody can quantify.
It will not give the industry's value, because the figure in circulation is a decade-old estimate its own author framed as an estimate.
What it will say is that the most-quoted dataset in this subject is a self-published marketing document, that it has been partially severed from the behaviour it once tracked, and that this is a more interesting fact about 2026 than any number that could have been printed in its place.