Nineteen games, zero players
The technical checklist was complete and it bought nothing. The real failures were keywords nobody searches for and a catalogue with no identity, and the ad rejection was a diagnosis rather than an insult.
On 13 August 2026 this site went live with nineteen original games. Every one of them was written here, in vanilla JavaScript, with no engine and no borrowed art. Two weeks later the analytics dashboard said the same thing it had said on day one: zero sessions. Not a handful. Not a slow start. Zero.
The ad network application was rejected in the same fortnight. The reason given was the one every publisher dreads because it sounds like an insult and is actually a diagnosis: not enough original content of value to users. Nineteen games I had personally built, and the verdict was that there was nothing here.
Both of those turn out to be the same failure, and it is worth writing down precisely, because the obvious explanations were all wrong.
The plumbing was never the problem
The first instinct after a launch with no traffic is to suspect the technical layer. That instinct was cheap to test and it came back clean. The sitemap was valid and submitted. The robots file allowed everything that mattered. Every page carried a self-referencing canonical URL, a description inside the length a search engine will actually display, structured data that parsed, and a share image at the right dimensions. Twenty-three URLs were indexable and the search console agreed they were indexable. Page weight was small, there were no render blockers, and the games loaded in well under a second.
It is worth saying this clearly because a lot of advice about launches begins and ends with that checklist. The checklist was complete and it bought nothing. Technical correctness is a floor, not a strategy — it removes reasons to be excluded, and removing reasons to be excluded is not the same as giving anyone a reason to arrive.
Failure one: the keywords were words nobody types
Every game page was optimised for a phrase describing the game. The trouble is that I had invented the games, so I had also invented the phrases. Duckling herding game. Neon dodge. These are accurate descriptions and they have essentially no search volume, because nobody wakes up wanting a duckling herding game. They want “free games” or “games to play at school” or a specific title they already heard of somewhere else.
The generic terms are owned. Poki, CrazyGames and itch sit on them with domain authority accumulated over a decade and catalogues in the tens of thousands. A new domain does not outrank them, this year or next year. That is not a pessimistic estimate; it is an arithmetic one. So the honest position was that search was never going to be the first channel, and every hour spent tuning page titles for it was an hour spent on the wrong problem.
Failure two: nineteen games, zero brand
The second failure is the one that took longer to see, because it does not show up in any tool. The games had nothing in common. A snake variant, a brick breaker, a reaction test, a duckling herder, a physics one-shot. Each was fine. None of them explained the others.
Play through one and leave, and there is no reason to remember the site. There is no character you would recognise again, no look you would pick out of a grid of thumbnails, no sentence a player could use to describe this place to somebody else. Nineteen games had been built and zero brands. In a market where discovery happens because someone shares a thing with a friend, a catalogue that cannot be described is a catalogue that cannot be shared.
Why the ad rejection was the same failure
The rejection looked unfair at first — the games are real, they are original, and they work. But the reviewer was not playing the games. The game files themselves carry a noindex tag, because the landing page around each game is the canonical URL. What a reviewer actually saw was twenty pages of near-identical structure: a title, a one-line tagline, three bullet points of instructions, a controls table, two questions in an FAQ. Roughly two hundred and fifty words each, in the same order, twenty times.
Read that as a stranger and it does not look like a studio. It looks like output from a template — which is the exact shape the low-value-content rule is written to catch. The complaint was never that the games were worthless. The complaint was that the site around them said almost nothing, and said it in the same voice twenty times over.
There was a second, smaller problem in the same direction: three sprite packs downloaded from itch.io were still in use across four games, and their licences had never been confirmed as permitting commercial use. Advertising had been switched on over pages carrying art whose rights had not been checked. That was fixed the direct way — the files were deleted and the games that used them were redrawn in code — but it should never have got as far as being live.
What changes
Three conclusions came out of this, and everything published on this site since has followed from them.
Fewer games, more of each. Nineteen mediocre games are worth less than three good ones, for every purpose at once: they do not persuade a reviewer, they do not survive a comparison with an established portal, and they give nobody anything to talk about. The count stopped being a metric.
One world instead of a catalogue. People follow characters, not genres. The newest game on the site, Feltling, is the first built with that in mind: a felt world with a seal and a cat in it, where the thing you remember is the seal rather than the mechanic. Whether that works is not yet known. It is at least a thing a player could describe to someone else in one sentence, which the previous nineteen were not.
The site has to say something. Pages that exist only to wrap a game in markup are not content, whatever the word count says. These dev notes are the other half of that answer — the part of this work that only I can write, because I am the person who made the mistake being described.
Zero sessions is a clean number to be handed. It removes any temptation to tell yourself the launch half-worked.