Choosing launch countries for acquisition: a low CPM is not enough. The usage density, media cost and regulation grid to decide where to launch an app or game in 2026.
"We launch first where the CPM is low, we learn, then we go after the big markets." I hear it at every app or game launch. And it is the best way to learn something false. A one-dollar CPM buys you impressions among people who will never pay, on creatives you will optimise for them, with retention curves that look nothing like those of your real markets. A launch country is not a cost. It is a dataset. And a false dataset costs more than any American CPM.
Launch countries: usage density before market size
Start with the big numbers, because they mislead. Newzoo's Global Games Market Report 2026 (25 August 2026) counts 3.7 billion players and $213.9 billion in spending. Asia-Pacific weighs 47% of the market with $100.7 billion for 1.94 billion players. North America, $56.9 billion for 251 million players. Europe, $38.5 billion for 475 million. Latin America, $9.3 billion for 394 million. Middle East and Africa, $8.5 billion for 640 million, the fastest-growing region at 10.3% a year.
Now do the division. A North American player spends about $227 a year. A European player, $81. An Asia-Pacific player, $52. A Latin American player, $24. A Middle Eastern or African player, $13. The ratio between the first and the last is 17. Newzoo adds that China and the US alone will account for 52% of global spending in 2026, and that gaming penetration has plateaued at 62% of the online population. There are no new players left to find. There are players who spend, and players who install.
Sensor Tower's State of Mobile 2026 (30 January 2026) says the same thing on the app side: India is the world's number one country for downloads, 25.5 billion installs in 2025, ahead of the US and Brazil. And the US is the number one country for spending, $59 billion. The country of installs and the country of money are not the same country. The usage density that matters to you is not "how many people play", it is "how many people play what you make, and how many pay for it". A competitive game, a niche app, a marketplace: each has its own density map, and it looks nothing like the world map.
A low CPM is a trap unless you divide it by value
The media cost gaps are real. Lebesgue's Meta CPM benchmark by country (July 2026, ecommerce data) gives a Facebook CPM of $16.08 in the US, $11.81 in the UK, $9.05 in Germany, $6.95 in France, $2.63 in Brazil, $1.36 in India. A one-to-twelve ratio. On installs, Adjust's Gaming App Insights Report 2026 (15 April 2026) gives a CPI of $1.68 in North America, $0.53 in Europe, $0.14 in Latin America, against a global average of $0.56, and all those prices rose 30 to 47% in a year.
But put the two tables side by side. The US CPM costs six times the Brazilian CPM. The North American player is worth almost ten times the Latin American player. On that basis, Brazil is not cheaper. It is more expensive, relative to what a player brings in. And the same Adjust report adds a figure few people look at: the ratio of paid installs to organic installs. It is 5.32 in Brazil, 4.28 in Turkey, 3.81 in India, against 3.53 in the US. In the "cheap" countries, you buy almost everything. Organic does not follow, because the usage density of your category is not there.
The only country metric worth anything is the cost of a user who pays, or who is still playing on day 30, relative to what that user is worth. Not CPM. Not CPI. We wrote it for scale-ups in general in the scale-up gap, and it is even truer when comparing countries: a global blended CAC always hides one country funding the hole in another.
Regulation is a media cost, not a legal footnote
Third column of the grid, and the one most often discovered too late. In Belgium, the Gaming Commission classified paid loot boxes as gambling back in 2018, and the major publishers preferred to remove the mechanic rather than take the risk. In the Netherlands, after a Council of State ruling in 2022, the government is now pushing for an EU-wide ban. In the UK, no gambling classification, but the ASA enforces loot box disclosure on store listings and parental spending caps. And Brussels is coming: the Digital Fairness Act sits in the Commission's 2026 work programme for the fourth quarter, with, according to Freshfields' analysis (November 2025), an obligation to display the real-world price of purchases made in virtual currency, restrictions or a ban on loot boxes for minors, and a 14-day withdrawal right on digital content.
This is not just a monetisation issue. It is an issue of buyable audience. Since 10 December 2025, Australia bans under-16s from social media on ten platforms, including Instagram, Facebook, Snapchat, TikTok, YouTube and Twitch, with fines of up to A$49.5 million. The result reported by Time on the day the law took effect: around 1.1 million accounts of 13-to-15-year-olds closed (440,000 at Snapchat, 350,000 on Instagram, 200,000 on TikTok, 150,000 on Facebook). If your target is 14 years old, Australia has no paid social left for you. Denmark is proposing the same under 15, Malaysia has announced it for 2026. A country where your audience can no longer be targeted on Meta and TikTok is not a country with "complex regulation". It is a country where the media plan does not exist, and you need to know it before, not after watching the platform reject the campaign.
The grid I use to rank countries
Three columns, each scored out of five. Density: the share of the population that practises your category, time spent, and spend per user, not the number of players. Cost: local CPM and CPI divided by the value of an active user, plus the paid-to-organic ratio. Regulation: what you are allowed to sell, who you are allowed to talk to, and on which platforms. You do not launch a country with a zero in one column, even with two fives next to it.
A country is several cities
Last point, and it is where most launch plans go off the rails: people reason in "countries" while media plays out in areas. A launch in Brazil is not a launch in Brazil. It is São Paulo, or it is nothing. Usage density is measured by city, and concentration is the only multiplier you control. For Deblock, we did exactly the opposite of a diluted national campaign: encircling metro stations with 100-metre mobile geo-targeting, more than 30 million impressions concentrated where the target is dense. The reasoning holds at the scale of a foreign country: you pick a country, then you pick its three cities, and you do not spend a euro anywhere else until retention has spoken.
And when you have to hold several markets at once, the discipline is the same: one grid, one threshold, one creative per market. For Sisley, we ran twelve markets with a 45% increase in ROAS, without damaging a luxury brand's image. Twelve markets is not twelve copies of the same plan. It is twelve different answers to the same grid. It is what we are working on right now with a video game client, a competitive gaming platform: ranking countries on these three columns before buying the first impression, then opening city by city, with geo-targeted social as the master channel and creatives tested locally, never translated from the market next door.
What I tell founders who "start with a cheap market"
You are not going to learn cheaper. You are going to learn wrong. Pick the country where your category is dense and where a user is worth something, even if that user costs four times more. Check what you are allowed to sell there and who you are allowed to talk to. Then launch three cities, not a country. The low CPM will come later, once you know what you are buying.
If you are preparing an international launch, bring your list of countries to our free audit: we run them through the density, cost and regulation grid together, and see which ones deserve the first impression. And to size the budget per market, the media mix simulator gives you the order of magnitude in a few minutes.
