We publish over a thousand browser games and record every session played on the portal. That produces a fairly unsentimental picture of what works in this format — one that does not always match what works on Steam, or on mobile, or in a game jam.
What follows is drawn from our own August 2026 measurement. None of it is a rule. All of it is a pattern we can see repeatedly, including in cases where we wish it were otherwise.
The tab is the whole constraint
Every difference between browser games and every other kind of game comes back to one fact: the player has invested nothing. There was no purchase, no install, no download bar to sit through. Closing your game costs exactly one keystroke and produces no regret.
On Steam, a player who paid twenty euros will push through a rough opening because sunk cost tells them to. In a tab, there is no sunk cost to work with. You have to earn the second minute with the first one, every single time.
Almost everything below is a consequence of that.
Immediacy beats depth
This is the clearest pattern in our data and the one developers find hardest to accept.
In our driving catalogue, crash sandboxes get opened more readily than racing games do — despite racing games generally being better designed. The reason is legibility: a crash simulator is comprehensible in about four seconds and satisfying in ten, while a racing game asks you to learn a track before it gives anything back. In a tab, the game that pays out immediately wins.
The same shows up elsewhere. Our most-played driving game is not a circuit racer, it is a traffic-dodging bike game — no track to memorise, tension from the first second. Our most-played 2-player game gives each player a single key.
None of this means shallow games are better. It means the depth has to be visible from the outside. Plants vs Zombies Fusion is the most played game on the entire portal and it is genuinely deep — but you understand the board in one screenshot. That combination is the target.
When a genre fights the format
We host nearly ninety idle and tycoon games, and together they are among the least opened categories on the site. This is worth understanding because it is not about quality — several are well made.
An idle game's core promise is that progress continues while you are away. That promise is what justifies a slow early game. But a closed browser tab has no background process, so for most browser idle games the game is simply paused. The bargain that makes the genre work quietly does not apply.
The idle games that do perform for us are the ones that noticed. Our best-performing example wraps the whole idle apparatus — upgrades, pets, rebirths — around an obstacle course you are actively moving through. You never wait. The accumulation happens underneath an activity instead of replacing one.
The general lesson: if your genre's core loop assumes a persistent session, you need to give the player something to do in the first two minutes, or port the assumption rather than the genre.
Load time is a design decision
Developers treat load time as an engineering detail to optimise at the end. In this format it is a design constraint on par with controls.
A cold load is the highest-risk moment your game has, because the player is looking at a progress bar having received nothing yet. Anything you can do to compress it — streaming assets, deferring what is not needed for the first level, cutting the initial bundle — buys you more players than any amount of polish applied later in the game.
If you can get to interactive in under five seconds on an average connection, you are ahead of most of our catalogue.
Your players are on worse machines than you
You are building on a development machine. A meaningful share of our audience is on school Chromebooks, library computers and five-year-old family laptops, frequently with a dozen other tabs open competing for the same GPU.
This has a concrete consequence: frame rate is worth more than fidelity. One of our best-performing shooters has deliberately plain visuals, and that is exactly why it works — the frame rate stays high and enemies stay readable, which matters far more in close-quarters play than texture resolution ever will.
Test on the worst hardware you can find rather than the best. If it holds up there, everything above it is free.
Launch week is not where the plays are
The instinct is to optimise for the launch spike. Our numbers say most of a game's lifetime plays arrive well afterwards, from search and from internal recommendations — provided there is a page substantial enough to rank and enough neighbouring games to be recommended from.
This is why we put roughly 1,500 words and ten FAQs on every game page in three languages: a thumbnail in a grid stops earning the day it stops being featured, while a real page keeps arriving in results months later.
From your side, the practical implication is to give a portal something to write about. A game with a documented control scheme, a clear genre, an explicable hook and a name people might actually search will out-earn a better game with none of those, because the second one is invisible to everything except the front page.
The short version
- Earn the second minute with the first. No install means no sunk cost to lean on.
- Make the depth visible from outside. Deep is fine; opaque is not.
- Check your genre survives the tab. If the loop assumes a persistent session, adapt it.
- Treat load time as design. Under five seconds to interactive.
- Test on bad hardware. Frame rate over fidelity, every time.
- Build for the long tail. Launch week is the smallest part of the total.
If you want this applied to your specific build rather than in the abstract, send it to us through the studios form. We play it the way a visitor would and write back with what we saw.
Frequently asked questions
Does this mean deep games do not work on the web?
No — it means depth has to be reachable. A deep game that is legible in the first thirty seconds does very well. A deep game that requires five minutes of setup before it becomes interesting loses most of its audience before it gets there, no matter how good minute six is.
How much does load time really matter?
More than any other single technical factor we can observe. A player who opened a tab on impulse has nothing invested and no install to justify. Every second of a cold load is a chance to close it.
Should I build for mobile as well?
It roughly doubles your reachable audience, so it is worth it if the controls survive the translation. What does not work is claiming touch support and shipping something unplayable with a thumb — that costs you more goodwill than having no mobile version at all.
Is it worth adding progression to a short game?
Often, yes. Games with something to unlock reliably get reopened more than games with a pure score. But bolting a progression system onto a game that does not need one adds friction to the first session, which is the session that decides everything.
Do these patterns apply to other portals too?
The format-level ones almost certainly do — the tab, the load time and the hardware constraints are properties of the browser, not of our site. The specific rankings are ours, drawn from our own August 2026 measurement, and any portal with a different audience would see different details.