Who ever actually paid independent game developers?

Every few years someone builds a place where independent developers can put their games and get paid. Most of these places are gone. A few worked well enough that the people who used them still talk about it.

This is the record: what was tried, what the split actually was, and what happened. We are writing it because we are about to build one, and the honest way to start is by reading the receipts of everyone who went first.

2005 to 2012: the sponsorship era

Before app stores, Flash games spread by being copied. A game would be uploaded to one portal, then embedded on hundreds of others. Nobody could stop that, so the business model was built on top of it instead.

Sponsorship meant a portal paid a developer to put its logo in the game, plus a "more games" link back. The portal was not buying the game. It was buying the traffic the game would carry as it spread across the internet. Distribution you cannot control becomes an asset the moment you get paid per copy of the brand it carries.

The early version was exclusive: a fixed sum, permanent, no in-game ads, no other licences. The developer got one cheque and gave up everything else.

FlashGameLicense changed the shape by pushing primary sponsorship: the developer keeps the right to run in-game ads and to license the game to other portals afterwards. The same game generated more total value, and the developer captured several streams from it. FGL was selling or licensing about three games a day by December 2011 and did close to 7,000 deals worth over 10 million dollars.

MochiAds did the other half. Developers embedded an ad unit inside the game itself, and got paid wherever the game ended up. The monetisation travelled with the file. Mochi Media raised 10 million dollars in 2008 and sold to Shanda Games for 80 million in 2010.

By the 2009 market survey, 58 percent of developers used around-game advertising, 43 percent used sponsorship, and a quarter earned from licensing. Twenty percent were making more than 1,000 dollars a month, which in 2009, for people making browser games alone in their bedrooms, was a real living.

Then Flash died, and all of it went with the runtime.

2007 to 2020: the walled portal

Kongregate launched in 2006 and did the thing that looked obviously correct: gather the games in one place, build an audience, share the money. Developers got between 25 and 50 percent of the ad revenue their game generated, and 70 percent of virtual goods sales through the platform currency.

Those are not bad terms. Kongregate still lost.

In July 2020, after layoffs, it stopped accepting new games and shut down its forums and chat. Not because the split was wrong, but because the players left. Developers went to Steam, to itch.io, to the App Store, and the audience followed. A portal is only worth its cut while it is where the people are.

Newgrounds, from the same world, is still alive, and how it survived is the interesting part. It kept an ad revenue share, paid out from 50 dollars, and then leaned on something else entirely: Supporter memberships. In March 2026 it raised new-supporter pricing from 3 to 5 dollars a month, and from 25 to 36 dollars a year, grandfathering everyone already subscribed. The stated goal is to go fully ad-free and push any surplus back into the revenue share.

Newgrounds outlived Kongregate by not depending on advertising alone.

2015 onward: the open store

itch.io did something no one else has been willing to copy. Since March 2015 the revenue split is set by the developer. The default is 10 percent, and a developer can move it anywhere between zero and 100.

The industry standard is 30 percent. Steam, the App Store and Google Play all take roughly that.

Giving away pricing power should be commercially insane. It worked, because it bought something a lower fixed rate cannot buy: developers believe itch.io is on their side, and act accordingly. Pay-what-you-want pricing sits on the same foundation, and buyers using it pay around 30 percent above the stated minimum on average. Several times a year, Creator Day waives the platform fee entirely.

itch.io now hosts over a million games. It is the largest catalogue of games in existence and almost nobody outside the scene has heard of it.

The platform economy: Roblox

Roblox is the biggest counter-argument to everything above, and also the clearest warning.

The money is enormous. Developers earned about 1.5 billion dollars through the Developer Exchange in 2025, up from roughly 923 million the year before, with more than 4 billion paid out since the programme began.

Then look at the distribution:

Twenty-five percent is the lowest effective share of any major distribution platform, and the payout curve is close to vertical. Roblox proves that a platform can pay out billions and still be a bad deal for the median participant. Both facts are true at once, and quoting either one alone is a way of lying with real numbers.

The modern web publishers

The Flash sponsorship model did not actually die. It came back with HTML5, and the two companies running it now are doing well.

Poki takes a straight 50/50 split with a condition that deserves more attention than it gets: if the player arrived through your own channel, you keep 100 percent. Bookmarks, your own social media, your community, search you earned: Poki takes nothing. It only takes half of the traffic it actually delivered. Poki passed a billion monthly plays in 2026 with more than 600 independent studios.

CrazyGames gives developers 60 percent of in-game advertising and 70 percent of in-game purchases, paying out over 100 euros. It reports more than 300 million gameplays a month across 50 million monthly players.

The reality check is in the earnings, not the percentages. A first deal is typically hundreds to low thousands of dollars a month. Top games on CrazyGames clear five figures monthly. The median is far below that.

And then, last year, someone rebuilt web rings

In March 2025 the Vibe Coding Game Jam added a rule that has more strategic content than any revenue split on this page.

Every game had to contain an exit portal. A ring you could walk, drive, swim, fly or sail into, which dropped you directly into another game in the jam.

The protocol is small and it works:

• the portal redirects to a shared endpoint that forwards the player onward
• GET parameters travel with the player: username, colour, speed
• ?ref= gives the receiving game a portal back to where the player came from
• ?portal=true tells the receiving game the player arrived through a portal, so it can skip the title screen and drop them straight in, mid-stride

And the incentive on top: the more players your game sends out, the more it receives back.

The 2026 edition ran 945 games, 242,212 players and around 12 million views on X. The comparison people reached for was the 1990s web ring, and it is the right one. A web ring was how small sites shared an audience before search engines decided who deserved one.

Note what this is. It is a distribution network with no store, no cut and no gatekeeper. Nobody takes a percentage, because nobody is processing a payment. It moves players, which is the thing every platform on this page was actually selling.

What worked

Four things, across twenty years and every era.

1. Monetisation that travels with the game. MochiAds paid wherever the file ended up. The game is the unit of distribution, not the store page. Every model that assumed players come to a central place eventually lost to the places players actually went.

2. Charging only for what you deliver. Poki takes half of the traffic it sent and nothing from the traffic you brought. This is the most honest term on this page and it is also good business, because it makes the platform impossible to resent.

3. Letting the developer set the rate. itch.io's default 10 percent, adjustable to zero, should not work. It does, because a platform that can be exploited and is not becomes the one people choose while nobody is forcing them to.

4. Reciprocity instead of aggregation. The Vibeverse gave games a reason to send players away: sending gets you receiving. No catalogue can produce that, because a catalogue's incentive is to keep you.

What did not work

1. Exclusive lock-in. The early Flash model bought a game outright for a fixed sum. It killed the developer's upside and got replaced by primary sponsorship as soon as an alternative existed.

2. Advertising alone on a walled portal. Kongregate's terms were fine. Its dependency was not. When the audience moved, the model had nothing else holding it up. Newgrounds survived the same decade because memberships gave it a second leg.

3. A platform currency with a bad exchange rate. Roblox pays out billions and leaves 85 percent of its developers under 100 dollars a month. A closed currency lets a platform set the real split far below the advertised one, and the median developer absorbs the difference.

4. Being a catalogue. This is the quiet one. A list of games is the easiest thing to build and the least defensible thing to own. Every directory in this field, including the current crop of vibe-coded game listings, has the same problem: there is no moat in a list. What is defensible is the thing that only exists because games are connected: traffic between them, identity across them, and a record of what happened.

What we take from it

We are not going to out-catalogue Poki. They have a billion plays a month and 600 studios, and that fight is over before it starts.

What none of them have is the thing the Vibeverse pointed at and did not finish. The portals move a player between games, but nothing survives the jump. No identity, no score, no history. You arrive in the next game as a stranger, every time. The username and colour ride along in the query string and are forgotten the moment you leave.

That gap is exactly the shape of what we already built. A score snippet that gives a player one identity and one ranking across every game that joins is the missing half of a portal network.

So the position is:

• speak the protocol that already exists rather than invent a competing one, because a second standard in a field this small helps nobody
• carry identity and score across the jump, which is what turns a hop into a journey
• take nothing from traffic we did not send, following Poki, because it is correct
• the game stays the developer's game. We are a layer, not a landlord.

Monetisation is the honest open question. Nobody earns anything here yet, and we are not going to pretend otherwise or invent a token to paper over it. What the record above suggests is that the models worth trying pay for movement rather than shelf space: something closer to Flash-era sponsorship, where a game earned by carrying players onward, than to a store taking thirty percent of a transaction that a store did not cause.

We will publish the numbers as we go, including the ones that make us look bad.

---

Sources

• Mochi Media Flash Games Market Survey 2009 and 2010; TechCrunch on Mochi Media's 10 million dollar raise (2008) and 80 million dollar sale to Shanda Games (2010)
• FlashGameLicense interview, GameDev.net; Dice.com on the Flash game licensing business model
• Kongregate support documentation on revenue share; GameSpot and Slashdot on the July 2020 submission freeze and layoffs
• Newgrounds Wiki on revenue sharing and Supporter status, including the March 2026 pricing change
• itch.io, "Introducing open revenue sharing" (March 2015) and itch.io documentation; Game Developer on the launch
• Roblox Creator Hub, Developer Exchange documentation; Statista on annual developer payouts; RoLearn on revenue share and earnings distribution
• Poki and CrazyGames developer terms as documented in 2026 web game monetisation guides
• 2025 Vibe Coding Game Jam rules and the Vibeverse portal specification, including the Three.js start and exit portal gist; vibej.am for the 2026 edition figures

Where a figure comes from a secondary guide rather than the platform itself, we say so rather than dressing it up as official.