How big is the gambling black market? It depends who is counting
Regulators, trade bodies and consultancies all publish a share for unlicensed gambling. For the same country and year the figures can sit 20 or 30 points apart, and the gap is not a mistake: it is the method. Here is every estimate we could trace to a primary document, side by side.
Every argument about gambling policy reaches the same number in the end. Raise a tax, cap a deposit or ban an advert, and someone warns that the black market will grow; leave the rules loose, and someone else insists the black market is small. The figure both sides reach for is the channelisation rate: the share of a country's gambling that happens with licensed operators. It sounds like a fact. It is an estimate, and the estimate depends on three choices that are rarely printed next to it.
One regulator, one report, three numbers
Start with the most careful regulator in the room. In June 2026 Sweden's Spelinspektionen published its channelisation report for 2025 and put the competitive online market at 84 per cent licensed, one point below 2024. That headline is an average of two indicators that disagree. Its player survey put 89 per cent of turnover with licensed sites; its estimate built from web traffic, weighted by what a visit is worth, put it at 78. For online casino the two readings sit 25 points apart, 93 per cent by survey and 68 by traffic, with 81 published. Count raw visits instead of money and only 54 per cent of online-casino traffic reaches licensed sites. Every one of those figures is in the same 25-page report, and the authority says plainly that its indicators are uncertain and rest on assumptions. Read as a black-market share, Swedish online casino is 7 per cent, 19 per cent, 32 per cent or 46 per cent. The regulator has not changed its mind between them. It has changed what it counts.
Players or money
The first choice is what gets counted. The Dutch Kansspelautoriteit publishes two channelisation rates every half-year. By players, 91 or 95 per cent of Dutch online gamblers stay with licensed sites, depending on whether someone who uses both legal and illegal sites counts as channelled; the authority calls both choices defensible and prints both. By money, its estimate of the licensed share of gross gaming revenue has slid to 53 per cent, from 56 the period before. Same market, same regulator, same months: 5 per cent of the people or 47 per cent of the money. The authority's own reading is that illegal players lose far more per head than legal ones, so a head-count will always look healthier than a cash-count. Norway's Lotteritilsynet says the same in different words: measured in players, channelling to the state monopoly edged up again in 2025; measured in money, foreign operators hold 28 to 33 per cent of the games they compete in, and the authority writes that money is the harder thing to count. Germany's regulator-commissioned survey found 20 per cent of provider mentions, 22 per cent of stakes and 23 per cent of losses on unlicensed sites, each figure a little larger than the last as the measure moves from people to cash.
Who paid for the count
The second choice is who commissioned the estimate. In November 2025 the Betting and Gaming Council, the British operators' trade body, published a PwC report warning that France had a 57 per cent black market, the Netherlands 37 per cent and Sweden 35 per cent. Those are not the regulators' numbers. They come from H2 Gambling Capital, a market-data firm, and PwC's own comparison page sets them beside the official ones: 65 per cent channelled in Sweden against the 82 it records for Spelinspektionen's 2024 estimate, 61 per cent in the Netherlands against the Kansspelautoriteit's 49. The report keeps the H2 figures, it says, for consistency across jurisdictions. Notice that the direction is not fixed: in Sweden the industry-published figure is 17 points gloomier than the regulator's, in the Netherlands 12 points rosier. In Germany the regulator's survey put 77 per cent of losses with licensed sites, while a University of Leipzig study paid for by the online-casino and sports-betting associations, using a metered panel of 25,000 consumers, put licensed usage at 51 per cent in March 2023. In France the regulator's own PwC study estimated illegal online revenue at €748 million to €1.5 billion, 5 to 11 per cent of the whole gambling market. Two years later the licensed operators' association updated the same PwC method, measured against online gambling alone, and announced that the illegal market had overtaken the legal one. Both can be accurate. They answer different questions.
The regulators who stopped publishing a number
The third choice is whether to publish a point at all. In November 2025 the Gambling Commission closed its series on Britain's illegal online market with the sentence that it was 'not yet in a position to make a robust and reliable estimate'. It dropped the survey approach because people cannot recall what they spent and often do not know whether a site is licensed, kept two traffic-based methods, and showed why they are fragile: with legal online yield at £6.9 billion, a half-point error in the channelisation rate moves the implied illegal market by £34.5 million. It then asked readers to be wary of third-party point estimates with no transparent method and no margin of error. Denmark's Spillemyndigheden, which reported 91 per cent of web traffic going to licensed sites in 2024 against a target of 85, left the rate out of its 2025 market review while it checks the figures, after an app estimate and a change in the traffic provider's algorithm made the series non-comparable. Germany's authority published its 77 per cent with the authors' warning that the panel is not representative of the population. Ontario, by contrast, runs the same Ipsos survey every spring and reports a rise to 91 per cent of online gamblers on regulated sites in 2026, from 84 the year before and against a pre-launch government estimate that 70 per cent of play was unregulated. It is a count of players, and says so.
What the number is for
None of this makes the black market imaginary. Every regulator above enforces against it, and the Dutch and Swedish series both drift downwards. It makes the number a lever rather than a reading. The British trade body cited France and Sweden to argue against a tax rise; the Dutch authority, asked whether its own tax rise pushed players offshore, answered that the decline in channelisation began before the rate went up and that too many rules changed at once to attribute it. Germany's states must evaluate their gambling treaty by the end of 2026 with channelisation as the central test, and the two sides arrive carrying 77 and 51. Brazil, which ended every online-betting authorisation in October 2026, will be measured next. Whoever quotes a share in those debates has already chosen a measure, a method and a sponsor, and the choice usually travels without its footnote.
How to read any black-market claim
Three questions strip a channelisation figure back to what it can support. What was counted: people, visits or money, and if money, stakes, gross revenue or losses. A head-count flatters the licensed market because heavy spenders go offshore; a visit-count punishes it because skin-betting and geoblocked sites pile up traffic that never becomes a bet. How it was measured: a survey inherits recall bias and the fact that many players cannot tell a licensed site from an unlicensed one; web traffic misses apps, which is where most licensed play now happens; search volume can be gamed, as the Dutch authority found when three operators' figures spiked; an operator's own estimate is an operator's own estimate. And who paid, with the scope they chose: online casino, all online gambling, or the whole market including lotteries. The honest figures above all come with a range, a caveat or both. When a share arrives as a single confident integer, the first thing it is missing is the method.
There is no single black-market number, and the regulators closest to the data say so. A channelisation rate is an estimate shaped by what was counted, how, and for whom; the same market can honestly read 93 per cent licensed by one measure and 54 by another in one report. Before repeating a figure, find its measure, its method and its sponsor. If the page that published it does not show all three, the number is an argument, not a measurement.
Understanding, not advice. This explains how a number works so you can read it clearly. It is not a system, and nothing here treats gambling as a way to make money — the maths favours the house. 18+.
A short, cited email when something in iGaming changes — plus one fundamental worth understanding. Independent, 18+, no hype.